Atak astmy
Rokowania, prognozy i postęp choroby

Ataki astmy stanowią poważne zagrożenie zdrowotne, a ich wczesne wykrycie i przewidywanie zaostrzeń są kluczowe dla skutecznego zarządzania chorobą. Najsilniejszym predyktorem przyszłych ataków jest historia wcześniejszych zaostrzeń, zwłaszcza w ciągu ostatniego roku. Inne istotne czynniki ryzyka to wiek podeszły, wcześniejsza wentylacja mechaniczna, obturacyjny bezdech senny, nadużywanie SABA, niekontrolowana astma, umiarkowana do ciężkiej depresja, eozynofilia, saturacja tlenem ≤90% (OR = 4,56; 95% CI 3,45-7,56; P≤0,001) oraz niskie wartości szczytowego przepływu wydechowego (PEF) po 1 godzinie leczenia (OR = 3,34; 95% CI 1,90-4,90; P≤0,001). Fenotypy zapalne krwi, takie jak wysoka liczba eozynofilów i niska liczba neutrofilów (HBE/LBN), również korelują z wyższym ryzykiem zaostrzeń, szczególnie u dzieci hospitalizowanych w sezonie zimowym i jesiennym. Epigenetyczne mechanizmy, w tym metylacja DNA i ekspresja mikroRNA, oraz czynniki genetyczne, takie jak regulacja chromatyny przez SMARCC1, SETD2, KMT2B i CHD8, odgrywają istotną rolę w patogenezie i prognozie astmy.

  1. Prognozowanie ataków astmy – wprowadzenie
  2. Czynniki ryzyka ataków astmy
    1. Genetyczne i epigenetyczne czynniki ryzyka
    2. Fenotypy astmy a ryzyko zaostrzenia
  3. Modele predykcyjne ataków astmy
    1. Metody uczenia maszynowego w przewidywaniu ataków astmy
    2. Wyzwania w opracowywaniu modeli predykcyjnych
    3. Innowacyjne podejścia do prognozowania ataków astmy
  4. Praktyczne zastosowania modeli predykcyjnych w zarządzaniu astmą
    1. Znaczenie wczesnej identyfikacji ryzyka
    2. Personalizacja leczenia astmy
  5. Metody oceny ryzyka ataku astmy
    1. Sezonowe wahania ryzyka ataku astmy
  6. Podsumowanie – prognoza ataków astmy
    1. Kolejne rozdziały

Prognozowanie ataków astmy – wprowadzenie

Ataki astmy (atak astmy) stanowią istotne zagrożenie dla zdrowia pacjentów, prowadząc do znacznego obniżenia jakości życia, zwiększonych kosztów opieki zdrowotnej i potencjalnie zagrażających życiu komplikacji. Wczesna identyfikacja pacjentów zagrożonych atakiem astmy oraz przewidywanie wystąpienia zaostrzeń są kluczowymi elementami skutecznego zarządzania chorobą.12 Ataki astmy w większości przypadków są możliwe do zapobieżenia, jeśli zostaną wcześnie wykryte i odpowiednio leczone.3

Modele prognostyczne mają ogromny potencjał w przewidywaniu zaostrzeń astmy, umożliwiając wczesną interwencję, co czyni je popularnym obszarem współczesnych badań.4 Zastosowanie algorytmów uczenia maszynowego (ML) oraz sztucznej inteligencji (AI) pozwala na opracowanie skutecznych narzędzi wczesnego ostrzegania, które analizują różne czynniki ryzyka i dostarczają informacji o prawdopodobieństwie wystąpienia niekorzystnego zdarzenia.5

Czynniki ryzyka ataków astmy

Identyfikacja głównych czynników ryzyka ataków astmy jest kluczowa dla skutecznej prognozy. Badania wykazały, że najsilniejszym predyktorem przyszłych ataków jest historia wcześniejszych zaostrzeń.67 Wiele modeli predykcyjnych wykorzystuje ten czynnik jako główny element prognozowania, gdyż osoby z astmą, które wcześniej doświadczyły ataku, są bardziej narażone na kolejne epizody.8

Do istotnych czynników ryzyka ciężkich zaostrzeń astmy należą:9

10

Badania wykazały również, że najważniejszymi predyktorami ciężkich zaostrzeń są: saturacja tlenem ≤90% na początku (OR = 4,56; 95% CI = 3,45-7,56; P≤0,001), wartość PEF po 1 godzinie leczenia (OR = 3,34; 95% CI = 1,90-4,90; P≤0,001) i niekontrolowana astma (OR = 3,33; 95% CI = 2,50-5,05; P≤0,001).11

Najdokładniejszym niezależnym predyktorem ciężkiego zaostrzenia astmy jest wartość PEF po 1 godzinie leczenia.12 Natomiast, jak wykazały badania, mimo że takie czynniki jak eozynofilia krwi, obniżony PEF, infekcje dolnych dróg oddechowych i młodszy wiek są istotnie związane ze zwiększonym ryzykiem przyszłych ataków astmy, to związki te są stosunkowo słabe i mogą nie być szczególnie pomocne w analizie ryzyka.13

Genetyczne i epigenetyczne czynniki ryzyka

Rodzinna atopia jest konsekwentnie uznawana za istotny predyktor astmy od dzieciństwa do dorosłości, przy czym dzieci rodziców z alergią wykazują 2-3 razy wyższe wskaźniki zachorowalności na astmę.14 Ponadto, badania podkreślają rolę czynników epigenetycznych, takich jak metylacja DNA, ekspresja mikroRNA i modyfikacja histonów w rozwoju astmy.15

Najnowsze badania zidentyfikowały również cztery kluczowe regulatory chromatyny (CR): SMARCC1, SETD2, KMT2B i CHD8, które mogą być wykorzystane do konstrukcji modelu nomogramowego do przewidywania prognozy pacjentów z ciężką astmą.16 KMT2B koduje enzym zaangażowany w metylację histonu H3 lizyny 4 (H3K4), a CHD8 koduje członka rodziny białek wiążących chromodomenę-helikazę-DNA, który odgrywa rolę w regulacji transkrypcji i remodelowaniu epigenetycznym.17

Fenotypy astmy a ryzyko zaostrzenia

Badania wykazały, że fenotypy astmy oparte na markerach zapalnych we krwi mają wpływ na ryzyko zaostrzeń. Fenotyp z wysoką liczbą eozynofilów i niską liczbą neutrofilów (HBE/LBN) wiązał się z wyższym ryzykiem zaostrzeń astmy wśród hospitalizowanych dzieci z astmą w zimie i jesieni, podczas gdy fenotyp z niską liczbą eozynofilów i niską liczbą neutrofilów (LBE/LBN) wiązał się z niższym ryzykiem w zimie, wiośnie i lecie.18 Eozynofile i neutrofile we krwi mogą mieć potencjalny wpływ na rozwój i ciężkość astmy u dzieci.19

Modele predykcyjne ataków astmy

Rozwój modeli predykcyjnych dla ataków astmy stanowi obszar intensywnych badań, szczególnie z wykorzystaniem uczenia maszynowego i sztucznej inteligencji. Celem tych modeli jest identyfikacja pacjentów z wysokim ryzykiem ataku oraz umożliwienie personalizacji leczenia i wczesnej interwencji.2021

Metody uczenia maszynowego w przewidywaniu ataków astmy

Badania wykazały, że modele oparte na uczeniu maszynowym mogą osiągać dobre wyniki w przewidywaniu zaostrzeń astmy. Metaanaliza 11 badań (23 modele predykcyjne) wykazała łączny wskaźnik pola pod krzywą charakterystyki operacyjnej odbiornika (AUROC) wynoszący 0,80 (95% CI 0,77-0,83).22 Modele te mogą przewidywać pacjentów z wysokim ryzykiem zaostrzenia od kilku dni do kilku lat wcześniej, co pomaga zidentyfikować osoby wymagające ściślejszego nadzoru.23

Wielkość próby ma kluczowe znaczenie dla wydajności modelu. Sugeruje to, że metody uczenia maszynowego będą preferowane dla modeli predykcyjnych tylko wtedy, gdy dostępny jest duży zbiór danych.24 Wśród stosowanych metod, algorytmy oparte na Gradient Boosted Decision Trees (GBDT) były najczęściej raportowane jako najlepiej działające.25

Badacze pracują nad stworzeniem zindywidualizowanego narzędzia oceny ryzyka dla klinicystów podstawowej opieki zdrowotnej, które pomoże w przewidywaniu ataków astmy w okresie 1, 4, 12, 26 i 52 tygodni. Wykorzystują do tego metodologie uczenia maszynowego, takie jak klasyfikatory bayesowskie, lasy losowe i maszyny wektorów nośnych, a także algorytmy zespołowe.26

Wyzwania w opracowywaniu modeli predykcyjnych

Pomimo zidentyfikowania wielu czynników ryzyka, identyfikacja osób z wysokim ryzykiem okazała się trudnym zadaniem.27 Większość modeli predykcyjnych wykazuje wysoką swoistość (prawidłowe przewidywanie niskiego ryzyka ataku u osób, które nie miały ataków), ale niską czułość (prawidłowe przewidywanie wysokiego ryzyka u osób, które ostatecznie miały ataki), co skutkuje mniej wiarygodną prognozą ryzyka dla pacjentów z grupy wysokiego ryzyka.28

Istnieje znaczna heterogeniczność w metodach uczenia maszynowego stosowanych w istniejących badaniach, co utrudnia znaczące porównanie. Przeglądy wskazują na kilka kluczowych wyzwań technicznych, które należy rozwiązać, aby postępować w kierunku wdrożenia klinicznego, takich jak problem nierównowagi klas, walidacja zewnętrzna, wyjaśnienie modelu i przestrzeganie wytycznych dotyczących raportowania w celu reprodukowalności modelu.29

Innowacyjne podejścia do prognozowania ataków astmy

Badanie DIGIPREDICT ma na celu identyfikację wczesnych cyfrowych markerów ataków astmy przy użyciu czujników wbudowanych w inteligentne urządzenia, w tym zegarki i inhalatory, oraz wykorzystanie zbiorów danych dotyczących zdrowia i środowiska oraz sztucznej inteligencji do opracowania modelu predykcji ryzyka, aby zapewnić wczesne, spersonalizowane ostrzeżenie o atakach astmy.30 Identyfikacja tych markerów umożliwi wczesne wykrywanie i zarządzanie atakami oraz poinformuje o rozwoju modelu predykcji ryzyka dla ataków astmy w oparciu o te cyfrowe markery.31

Innym innowacyjnym podejściem jest adaptacyjna struktura inteligentnego eZdrowia, która nie tylko przewiduje ataki astmy, ale również wykorzystuje dane przestrzenne do zapewnienia bezpiecznej trasy, która odsuwa pacjenta od wszelkich czynników wyzwalających astmę.32 System ten wykazał imponującą dokładność 98% w przewidywaniu ataków astmy z odwołaniem 96%.33 Jest to podejście proaktywne, pozwalające pacjentom na podjęcie środków zapobiegawczych przed wystąpieniem ataku astmy.34

Praktyczne zastosowania modeli predykcyjnych w zarządzaniu astmą

Modele predykcyjne dla ataków astmy mają potencjał do zrewolucjonizowania zarządzania tą chorobą poprzez umożliwienie wczesnej interwencji i personalizacji opieki.35 Takie narzędzia predykcyjne mogą znacznie poprawić wyniki pacjentów, umożliwiając wczesną interwencję i dostosowane strategie zarządzania dopasowane do indywidualnych profili ryzyka.36

Pomimo postępów w ML i AutoML dla zastosowań w opiece zdrowotnej, większość modeli predykcyjnych dla astmy została zwalidowana w izolowanych zestawach danych, co ogranicza ich możliwość generalizacji. Modele, które można uogólnić, są niezbędne, aby zapewnić spójną wydajność w różnych populacjach i warunkach klinicznych.37

Znaczenie wczesnej identyfikacji ryzyka

Wczesna identyfikacja świszczącego oddechu u dzieci poniżej piątego roku życia może dostarczyć cennych informacji rodzicom i personelowi medycznemu oraz pomóc we wczesnej stratyfikacji i ścisłym monitorowaniu pacjentów zagrożonych astmą.38 Modele predykcji astmy dziecięcej są pomocne w identyfikacji prawdopodobnych przyszłych pacjentów z astmą z grup wysokiego ryzyka; dzieci w wieku przedszkolnym, u których rozwijają się objawy, mogą odnieść korzyści z wczesnej diagnozy i interwencji.39

Według badania populacji będącej częścią FinEsS (Finlandia-Estonia-Szwecja), mediana wieku dla diagnozy astmy alergicznej wynosiła 19 lat, a dla astmy niealergicznej – 35 lat.40 Zapadalność na astmę alergiczną była wysoka w grupie wiekowej 0-9 lat (1,8/1000/rok) i niższa w grupie wiekowej 50-59 lat (0,6/1000/rok).41

Personalizacja leczenia astmy

Fenotyp alergiczny związany z astmą wykazuje różnorodne cechy, na które wpływa złożona interakcja czynników środowiskowych, genetycznych i psychospołecznych.42 Modele predykcyjne dla astmy dziecięcej sprawdziły się w rozpoznawaniu przyszłych astmatyków w grupach pacjentów wysokiego ryzyka poprzez ich stosowanie w okresie przedszkolnym, który jest kluczowym okresem dla rozwoju układu odpornościowego i wzrostu płuc.43

Obecna wiedza na temat epigenetyki, wykorzystanie biomarkerów i różnych typów algorytmów w przewidywaniu astmy u dzieci daje możliwość poprawy dokładności tych narzędzi diagnostycznych.44

Metody oceny ryzyka ataku astmy

Badania wykazały, że poziomy FEV1, FEV1/VC i PD20FEV1 poniżej wartości granicznych są wysoce predykcyjne dla kolejnego zaostrzenia astmy.45 Obecność dwóch lub trzech czynników ryzyka wiązała się ze znacznie wyższym ryzykiem zaostrzenia.46 Przy 2 z 3 czynników ryzyka, iloraz szans (OR) dla zaostrzenia astmy wzrósł do 5,25 i osiągnął 11,3 przy 3 czynnikach ryzyka.47

Sezonowe wahania ryzyka ataku astmy

Badania zidentyfikowały związki między fenotypami zapalnymi krwi w astmie a wzrostem zaostrzeń astmy u hospitalizowanych dzieci z astmą przed pandemią COVID-19 w sezonie zimowym i jesiennym.48 Fenotyp HBE/LBN był związany ze zwiększonymi zaostrzeniami astmy wśród hospitalizowanych dzieci z astmą w sezonie zimowym i jesiennym.49

Podsumowanie – prognoza ataków astmy

Przewidywanie ataków astmy stanowi kluczowy element efektywnego zarządzania tą chorobą. Najsilniejszym predyktorem przyszłych ataków jest historia wcześniejszych zaostrzeń, szczególnie w ciągu ostatniego roku.50 Modele oparte na uczeniu maszynowym wykazują obiecujące wyniki w przewidywaniu ryzyka zaostrzenia, przy czym metaanaliza wykazała łączny wskaźnik AUROC wynoszący 0,80.51

Istotne czynniki ryzyka obejmują obniżoną saturację tlenem (≤90%), niskie wartości szczytowego przepływu wydechowego, niekontrolowaną astmę, eozynofilię krwi oraz fenotypy zapalne.5253 Innowacyjne podejścia, takie jak wykorzystanie cyfrowych biomarkerów i inteligentnych urządzeń, mają potencjał do dalszej poprawy dokładności prognozowania.5455

Wczesna identyfikacja pacjentów z wysokim ryzykiem ataków astmy umożliwia personalizację leczenia i wczesną interwencję, co może znacząco poprawić wyniki kliniczne i jakość życia pacjentów z astmą.5657 Dalszy rozwój i walidacja modeli predykcyjnych, w tym zastosowanie uczenia maszynowego i sztucznej inteligencji, stanowi obiecujący kierunek badań nad astmą.5859

Kolejne rozdziały

Zapraszamy do dalszego czytania naszego leksykonu.

Wybierz kolejny rozdział z menu poniżej, aby otworzyć nową podstronę kompedium wiedzy i uzyskać szczegółowe informację o leku, substancji lub chorobie.

  1. 10.04.2026
  2. www.leksykon.com.pl

Materiały źródłowe

  • #1 Acute severe asthma in emergency department: clinical characteristics, risk factors, and predictors for poor outcome | The Egyptian Journal of Bronchology | Full Text
    https://ejb.springeropen.com/articles/10.1186/s43168-022-00160-8
    Severe asthma exacerbations can be predicted by old age, previous history of mechanical ventilation, obstructive sleep apnea, overuse of SABA, uncontrolled asthma, moderate to severe depression, eosinophilia, SO2 90%, and low peak expiratory flow rates. […] The most important predictors for severe exacerbations were SO2 90% at baseline (OR = 4.56; 95% CI = 3.457.56; P0.001), PEFR after 1 h (OR= 3.34; 95%CI = 1.904.90; P0.001), and uncontrolled asthma (OR= 3.33; 95%CI = 2.505.05; P0.001). […] Based on the current study, the predictors of hospitalization were old age (OR= 1.11; 95%CI= 1.092.11; P0.001), uncontrolled asthma (OR= 2.34; 95%CI= 2.014.40; P0.001), PEFR after 1 h (OR= 4.44; 95%CI= 3.247.68; P0.001), and SO2 90% at baseline (OR= 5.67; 95%CI= 3.988.50; P0.001). […] The most accurate independent predictor for severe asthma exacerbation is the PEFR value after 1 h of treatment.
  • #2 DIGIPREDICT: physiological, behavioural and environmental predictors of asthma attacks—a prospective observational study using digital markers and artificial intelligence—study protocol | BMJ Open Respiratory Research
    https://bmjopenrespres.bmj.com/content/11/1/e002275
    Asthma attacks are a leading cause of morbidity and mortality but are preventable in most if detected and treated promptly. […] The aim of this study DIGIPREDICT is to identify early digital markers of asthma attacks using sensors embedded in smart devices including watches and inhalers, and leverage health and environmental datasets and artificial intelligence, to develop a risk prediction model to provide an early, personalised warning of asthma attacks. […] Identifying these markers will enable early detection and management of attacks and inform the development of a risk prediction model for asthma attacks based on these digital markers. […] The aim of this study is to prospectively collect real-time data on digital biomarkers and other parameters that are associated with asthma attack risk, using digital technologies and wireless sensors, and combine this information with retrospective data from NZ health and environmental datasets to improve prediction of asthma attack risk. […] We hypothesise that there will be clear digital markers that will be associated with asthma attacks, and that their inclusion in a risk prediction model will improve the predictive ability of the models for asthma attacks.
  • #3 DIGIPREDICT: physiological, behavioural and environmental predictors of asthma attacks—a prospective observational study using digital markers and artificial intelligence—study protocol | BMJ Open Respiratory Research
    https://bmjopenrespres.bmj.com/content/11/1/e002275
    Asthma attacks are a leading cause of morbidity and mortality but are preventable in most if detected and treated promptly. […] The aim of this study DIGIPREDICT is to identify early digital markers of asthma attacks using sensors embedded in smart devices including watches and inhalers, and leverage health and environmental datasets and artificial intelligence, to develop a risk prediction model to provide an early, personalised warning of asthma attacks. […] Identifying these markers will enable early detection and management of attacks and inform the development of a risk prediction model for asthma attacks based on these digital markers. […] The aim of this study is to prospectively collect real-time data on digital biomarkers and other parameters that are associated with asthma attack risk, using digital technologies and wireless sensors, and combine this information with retrospective data from NZ health and environmental datasets to improve prediction of asthma attack risk. […] We hypothesise that there will be clear digital markers that will be associated with asthma attacks, and that their inclusion in a risk prediction model will improve the predictive ability of the models for asthma attacks.
  • #4 Primary Care Asthma Attack Prediction Models for Adults | JAA
    https://www.dovepress.com/primary-care-asthma-attack-prediction-models-for-adults-a-systematic-r-peer-reviewed-fulltext-article-JAA
    Prognostic models hold great potential for predicting asthma exacerbations, providing opportunities for early intervention, and are a popular area of current research. […] This systematic review aimed to identify novel predictive models of asthma attacks in adults and compare differences in construction related to populations, outcome definitions, prediction time horizons, algorithms, validation, and performance estimation. […] The outcome of the prediction model was the onset of an asthma attack, and studies that only reported other asthma-related events (such as post-asthma attack hospital discharge) or statuses (such as uncontrolled asthma or asthma severity) were excluded. […] The predictive performance is heavily influenced by the study design, including the population, the outcome definition, the algorithm, and the model validation procedures. Identifying the most clinically meaningful model characteristics is necessary to enable a best model to be identified and highlight routes for future development. This will boost likelihood of successful translation, adoption, and implementation at scale of clinical prediction models, and to bring benefits to patients.
  • #5 JMIR AI – Machine Learning–Based Asthma Attack Prediction Models From Routinely Collected Electronic Health Records: Systematic Scoping Review
    https://ai.jmir.org/2023/1/e46717
    Asthma attacks occur particularly in those with poorly controlled diseases. An asthma attack is a sudden or gradual deterioration of asthma symptoms that can have a major influence on a patients quality of life. Such attacks can be life-threatening and necessitate rapid medical attention, such as an accident and emergency department visit or hospitalization, and can even lead to mortality. Early warning tools to predict asthma attacks offer the opportunity to provide timely treatments and, thereby, minimize the risk of serious outcomes. […] Machine learning (ML) offers the potential to develop an early warning tool that takes different risk factors as input and then outputs the probability of an adverse outcome. The predictive performance of this method may be inferior to more advanced ML methods, especially for relatively high-dimensional data with complex and nonlinear relationships between the variables.
  • #6 Applying UK real-world primary care data to predict asthma attacks in 3776 well-characterised children: a retrospective cohort study | npj Primary Care Respiratory Medicine
    https://www.nature.com/articles/s41533-018-0095-5
    Current understanding of risk factors for asthma attacks in children is based on studies of small but well-characterised populations or pharmaco-epidemiology studies of large but poorly characterised populations. […] In this large population, several factors were associated with a future asthma attack, but a past history of attacks was most strongly associated with future attacks. […] Our results indicate that, of all the outcomes collected in this large study, a past asthma attack (and especially two attacks) is likely to be the best method to identify children who might benefit from a stratified intervention aimed at reducing their risk for future asthma attacks. […] Factors other than a past history of attacks which were associated with increased risk for future asthma attacks included blood eosinophilia, reduced PEF, lower respiratory tract infection and younger age, and although these associations were highly significant, they were weakly related to risk for asthma attacks and therefore not likely to be particularly helpful in risk analysis. […] In summary, we find that a past history of asthma attacks is the best predictor of future attacks, and that blood eosinophilia and reduced PEF do not add substantially to predicting attacks.
  • #7 Predicting asthma attacks in primary care: protocol for developing a machine learning-based prediction model | BMJ Open
    https://bmjopen.bmj.com/content/9/7/e028375
    While it might seem intuitive that the patients with most severe daily symptoms exhibit greater risk of severe morbidity and mortality, research suggests that these symptoms may be a suboptimal clinical marker of asthma attack risk. […] Indeed, some people with asthma are more prone to asthma attacks than others, with asthma attack history being the strongest risk factor for future asthma attacks. […] Despite the identification of many risk factors, identifying high-risk individuals has proven a challenging task. […] As such, most prediction models report high specificity (correctly predicting low attack risk to those who did not have attacks), but low sensitivity (correctly predicting high risk in those who did go on to have attacks), which results in less reliable risk prediction for patients at high risk.
  • #8 Predicting asthma attacks in primary care: protocol for developing a machine learning-based prediction model | BMJ Open
    https://bmjopen.bmj.com/content/9/7/e028375
    While it might seem intuitive that the patients with most severe daily symptoms exhibit greater risk of severe morbidity and mortality, research suggests that these symptoms may be a suboptimal clinical marker of asthma attack risk. […] Indeed, some people with asthma are more prone to asthma attacks than others, with asthma attack history being the strongest risk factor for future asthma attacks. […] Despite the identification of many risk factors, identifying high-risk individuals has proven a challenging task. […] As such, most prediction models report high specificity (correctly predicting low attack risk to those who did not have attacks), but low sensitivity (correctly predicting high risk in those who did go on to have attacks), which results in less reliable risk prediction for patients at high risk.
  • #9 Acute severe asthma in emergency department: clinical characteristics, risk factors, and predictors for poor outcome | The Egyptian Journal of Bronchology | Full Text
    https://ejb.springeropen.com/articles/10.1186/s43168-022-00160-8
    Severe asthma exacerbations can be predicted by old age, previous history of mechanical ventilation, obstructive sleep apnea, overuse of SABA, uncontrolled asthma, moderate to severe depression, eosinophilia, SO2 90%, and low peak expiratory flow rates. […] The most important predictors for severe exacerbations were SO2 90% at baseline (OR = 4.56; 95% CI = 3.457.56; P0.001), PEFR after 1 h (OR= 3.34; 95%CI = 1.904.90; P0.001), and uncontrolled asthma (OR= 3.33; 95%CI = 2.505.05; P0.001). […] Based on the current study, the predictors of hospitalization were old age (OR= 1.11; 95%CI= 1.092.11; P0.001), uncontrolled asthma (OR= 2.34; 95%CI= 2.014.40; P0.001), PEFR after 1 h (OR= 4.44; 95%CI= 3.247.68; P0.001), and SO2 90% at baseline (OR= 5.67; 95%CI= 3.988.50; P0.001). […] The most accurate independent predictor for severe asthma exacerbation is the PEFR value after 1 h of treatment.
  • #10 Acute severe asthma in emergency department: clinical characteristics, risk factors, and predictors for poor outcome | The Egyptian Journal of Bronchology | Full Text
    https://ejb.springeropen.com/articles/10.1186/s43168-022-00160-8
    Severe asthma exacerbations can be predicted by old age, previous history of mechanical ventilation, obstructive sleep apnea, overuse of SABA, uncontrolled asthma, moderate to severe depression, eosinophilia, SO2 90%, and low peak expiratory flow rates. […] The most important predictors for severe exacerbations were SO2 90% at baseline (OR = 4.56; 95% CI = 3.457.56; P0.001), PEFR after 1 h (OR= 3.34; 95%CI = 1.904.90; P0.001), and uncontrolled asthma (OR= 3.33; 95%CI = 2.505.05; P0.001). […] Based on the current study, the predictors of hospitalization were old age (OR= 1.11; 95%CI= 1.092.11; P0.001), uncontrolled asthma (OR= 2.34; 95%CI= 2.014.40; P0.001), PEFR after 1 h (OR= 4.44; 95%CI= 3.247.68; P0.001), and SO2 90% at baseline (OR= 5.67; 95%CI= 3.988.50; P0.001). […] The most accurate independent predictor for severe asthma exacerbation is the PEFR value after 1 h of treatment.
  • #11 Acute severe asthma in emergency department: clinical characteristics, risk factors, and predictors for poor outcome | The Egyptian Journal of Bronchology | Full Text
    https://ejb.springeropen.com/articles/10.1186/s43168-022-00160-8
    Severe asthma exacerbations can be predicted by old age, previous history of mechanical ventilation, obstructive sleep apnea, overuse of SABA, uncontrolled asthma, moderate to severe depression, eosinophilia, SO2 90%, and low peak expiratory flow rates. […] The most important predictors for severe exacerbations were SO2 90% at baseline (OR = 4.56; 95% CI = 3.457.56; P0.001), PEFR after 1 h (OR= 3.34; 95%CI = 1.904.90; P0.001), and uncontrolled asthma (OR= 3.33; 95%CI = 2.505.05; P0.001). […] Based on the current study, the predictors of hospitalization were old age (OR= 1.11; 95%CI= 1.092.11; P0.001), uncontrolled asthma (OR= 2.34; 95%CI= 2.014.40; P0.001), PEFR after 1 h (OR= 4.44; 95%CI= 3.247.68; P0.001), and SO2 90% at baseline (OR= 5.67; 95%CI= 3.988.50; P0.001). […] The most accurate independent predictor for severe asthma exacerbation is the PEFR value after 1 h of treatment.
  • #12 Acute severe asthma in emergency department: clinical characteristics, risk factors, and predictors for poor outcome | The Egyptian Journal of Bronchology | Full Text
    https://ejb.springeropen.com/articles/10.1186/s43168-022-00160-8
    Severe asthma exacerbations can be predicted by old age, previous history of mechanical ventilation, obstructive sleep apnea, overuse of SABA, uncontrolled asthma, moderate to severe depression, eosinophilia, SO2 90%, and low peak expiratory flow rates. […] The most important predictors for severe exacerbations were SO2 90% at baseline (OR = 4.56; 95% CI = 3.457.56; P0.001), PEFR after 1 h (OR= 3.34; 95%CI = 1.904.90; P0.001), and uncontrolled asthma (OR= 3.33; 95%CI = 2.505.05; P0.001). […] Based on the current study, the predictors of hospitalization were old age (OR= 1.11; 95%CI= 1.092.11; P0.001), uncontrolled asthma (OR= 2.34; 95%CI= 2.014.40; P0.001), PEFR after 1 h (OR= 4.44; 95%CI= 3.247.68; P0.001), and SO2 90% at baseline (OR= 5.67; 95%CI= 3.988.50; P0.001). […] The most accurate independent predictor for severe asthma exacerbation is the PEFR value after 1 h of treatment.
  • #13 Applying UK real-world primary care data to predict asthma attacks in 3776 well-characterised children: a retrospective cohort study | npj Primary Care Respiratory Medicine
    https://www.nature.com/articles/s41533-018-0095-5
    Current understanding of risk factors for asthma attacks in children is based on studies of small but well-characterised populations or pharmaco-epidemiology studies of large but poorly characterised populations. […] In this large population, several factors were associated with a future asthma attack, but a past history of attacks was most strongly associated with future attacks. […] Our results indicate that, of all the outcomes collected in this large study, a past asthma attack (and especially two attacks) is likely to be the best method to identify children who might benefit from a stratified intervention aimed at reducing their risk for future asthma attacks. […] Factors other than a past history of attacks which were associated with increased risk for future asthma attacks included blood eosinophilia, reduced PEF, lower respiratory tract infection and younger age, and although these associations were highly significant, they were weakly related to risk for asthma attacks and therefore not likely to be particularly helpful in risk analysis. […] In summary, we find that a past history of asthma attacks is the best predictor of future attacks, and that blood eosinophilia and reduced PEF do not add substantially to predicting attacks.
  • #14 Early Prediction of Asthma
    https://pmc.ncbi.nlm.nih.gov/articles/PMC10455492/
    According to a study of a population that was part of FinEsS (Finland-Estonia-Sweden), the median age for the diagnosis of allergic asthma was 19 years, and that for non-allergic asthma was 35 years. […] The incidence of allergic asthma was high in the 0-9 age group (1.8/1000/year) and lower in the 50-59 age group (0.6/1000/year). […] The allergic phenotype associated with asthma exhibits a diverse range of characteristics that are influenced by a complex interplay of environmental, genetic, and psychosocial factors. […] Familial atopy is consistently recognized as a significant predictor of asthma from childhood to adulthood, with the children of allergic parents manifesting rates of asthma that are two to three times higher. […] Prediction models for childhood asthma have been proven to be functional in recognizing future asthmatics in high-risk groups of patients through their use in the preschool period, which is a crucial period for immune development and lung growth. […] Current knowledge of epigenetics, the use of biomarkers, and different types of algorithms in the prediction of asthma in children provide an opportunity to improve the accuracy of these diagnostic tools.
  • #15 Early Prediction of Asthma
    https://pmc.ncbi.nlm.nih.gov/articles/PMC10455492/
    The clinical manifestations of asthma in children are highly variable, are associated with different molecular and cellular mechanisms, and are characterized by common symptoms that may diversify in frequency and intensity throughout life. […] It is essential to promptly identify patients at high risk of developing asthma by using different prediction models. […] This review also highlights the epigenetic factors that are involved, such as DNA methylation and asthma risk, microRNA expression, and histone modification. […] The early identification of wheezing in children younger than five years old can provide valuable information to parents and medical professionals and aid in the early stratification and close monitoring of patients at risk of asthma. […] Prediction models for childhood asthma are helpful for identifying likely future asthma patients from high-risk groups; children in preschool who develop symptoms could benefit from early diagnosis and intervention.
  • #16 A prognosis prediction chromatin regulator signature for patients with severe asthma | Allergy, Asthma & Clinical Immunology | Full Text
    https://aacijournal.biomedcentral.com/articles/10.1186/s13223-023-00796-1
    Severe asthma imposes a physical and economic burden on both patients and society. […] The nomogram constructed using the four CRs, SMARCC1, SETD2, KMT2B, and CHD8, may be a useful tool for predicting the prognosis of patients with severe asthma. […] To further investigate the relationship between CRs and prognosis in patients with severe asthma, we constructed a prognostic prediction model using the four identified key CRs: SMARCC1, CHD8, SETD2, and KMT2B. The model showed a good predictive performance for prognosis. […] These four genes, SMARCC1, SETD2, KMT2B, and CHD8, were used to construct a nomogram model for predicting the prognosis of patients with severe asthma. […] The results showed a negative correlation between SMARCC1 and T helper cells and a negative correlation between SETD2 and APC co-inhibition, para-inflammation, treg, and type I IFN responses.
  • #17 A prognosis prediction chromatin regulator signature for patients with severe asthma | Allergy, Asthma & Clinical Immunology | Full Text
    https://aacijournal.biomedcentral.com/articles/10.1186/s13223-023-00796-1
    The model also involved KMT2B, which encodes an enzyme involved in histone H3 lysine 4 (H3K4) methylation, and CHD8, which encodes for a member of the chromodomain-helicase-DNA binding protein family that has been reported to play a role in transcriptional regulation, epigenetic remodeling, and other processes. […] The results of this study provide new insights into the mechanisms underlying CRs in severe asthma.
  • #18 Seasonal variation of pediatric asthma exacerbations and its association with asthma phenotypes | Pediatric Research
    https://www.nature.com/articles/s41390-025-04073-2
    HBE/LBN phenotype had a higher risk of asthma exacerbations among hospitalized pediatric asthma patients in the winter and autumn, while LBE/LBN phenotype had a lower risk in the winter, spring, and summer. […] Blood eosinophils and neutrophils have been indicated to have a potential influence on pediatric asthma development and severity. […] HBE/LBN phenotype was associated with increased asthma exacerbations among hospitalized pediatric asthma patients during winter and autumn. […] Eosinophil and neutrophil predominance exhibited a higher influence on pediatric asthma exacerbations. […] We observed that asthma phenotypes were associated with an increase in asthma exacerbations in hospitalized pediatric asthma patients. […] Our study further identified that asthma blood inflammatory phenotypes were associated with an increase in asthma exacerbations of hospitalized pediatric asthma patients during winter and autumn.
  • #19 Seasonal variation of pediatric asthma exacerbations and its association with asthma phenotypes | Pediatric Research
    https://www.nature.com/articles/s41390-025-04073-2
    HBE/LBN phenotype had a higher risk of asthma exacerbations among hospitalized pediatric asthma patients in the winter and autumn, while LBE/LBN phenotype had a lower risk in the winter, spring, and summer. […] Blood eosinophils and neutrophils have been indicated to have a potential influence on pediatric asthma development and severity. […] HBE/LBN phenotype was associated with increased asthma exacerbations among hospitalized pediatric asthma patients during winter and autumn. […] Eosinophil and neutrophil predominance exhibited a higher influence on pediatric asthma exacerbations. […] We observed that asthma phenotypes were associated with an increase in asthma exacerbations in hospitalized pediatric asthma patients. […] Our study further identified that asthma blood inflammatory phenotypes were associated with an increase in asthma exacerbations of hospitalized pediatric asthma patients during winter and autumn.
  • #20 Primary Care Asthma Attack Prediction Models for Adults | JAA
    https://www.dovepress.com/primary-care-asthma-attack-prediction-models-for-adults-a-systematic-r-peer-reviewed-fulltext-article-JAA
    Prognostic models hold great potential for predicting asthma exacerbations, providing opportunities for early intervention, and are a popular area of current research. […] This systematic review aimed to identify novel predictive models of asthma attacks in adults and compare differences in construction related to populations, outcome definitions, prediction time horizons, algorithms, validation, and performance estimation. […] The outcome of the prediction model was the onset of an asthma attack, and studies that only reported other asthma-related events (such as post-asthma attack hospital discharge) or statuses (such as uncontrolled asthma or asthma severity) were excluded. […] The predictive performance is heavily influenced by the study design, including the population, the outcome definition, the algorithm, and the model validation procedures. Identifying the most clinically meaningful model characteristics is necessary to enable a best model to be identified and highlight routes for future development. This will boost likelihood of successful translation, adoption, and implementation at scale of clinical prediction models, and to bring benefits to patients.
  • #21 Early Prediction of Asthma
    https://pmc.ncbi.nlm.nih.gov/articles/PMC10455492/
    According to a study of a population that was part of FinEsS (Finland-Estonia-Sweden), the median age for the diagnosis of allergic asthma was 19 years, and that for non-allergic asthma was 35 years. […] The incidence of allergic asthma was high in the 0-9 age group (1.8/1000/year) and lower in the 50-59 age group (0.6/1000/year). […] The allergic phenotype associated with asthma exhibits a diverse range of characteristics that are influenced by a complex interplay of environmental, genetic, and psychosocial factors. […] Familial atopy is consistently recognized as a significant predictor of asthma from childhood to adulthood, with the children of allergic parents manifesting rates of asthma that are two to three times higher. […] Prediction models for childhood asthma have been proven to be functional in recognizing future asthmatics in high-risk groups of patients through their use in the preschool period, which is a crucial period for immune development and lung growth. […] Current knowledge of epigenetics, the use of biomarkers, and different types of algorithms in the prediction of asthma in children provide an opportunity to improve the accuracy of these diagnostic tools.
  • #22 Machine learning for prediction of asthma exacerbations among asthmatic patients: a systematic review and meta-analysis | BMC Pulmonary Medicine | Full Text
    https://bmcpulmmed.biomedcentral.com/articles/10.1186/s12890-023-02570-w
    Asthma exacerbations reduce the patients quality of life and are also responsible for significant disease burdens and economic costs. […] This systematic review and meta-analysis aimed to identify the prediction performance of ML-based prediction models for asthma exacerbations and address the uncertainty of whether modern ML methods could become an alternative option to predict asthma exacerbations. […] The overall pooled area under the curve of the receiver operating characteristics (AUROC) of 11 studies (23 prediction models) was 0.80 (95% CI 0.770.83). […] This study identified that ML was the potential tool to achieve great performance in predicting asthma exacerbations. […] Future studies should focus on improving the generalization ability and practicability, thus driving the application of these models in clinical practice.
  • #23 Machine learning for prediction of asthma exacerbations among asthmatic patients: a systematic review and meta-analysis | BMC Pulmonary Medicine | Full Text
    https://bmcpulmmed.biomedcentral.com/articles/10.1186/s12890-023-02570-w
    The overall pooled AUROC (0.8, 95% CI 0.760.83) and DOR (7.02, 95% CI 5.209.47) indicated that ML-based prediction models for asthma exacerbation could achieve good discrimination. […] ML prediction models could forecast patients at high risk of exacerbation from several days to years, helping identify patients needing closer management. […] The sample size is crucial for model performance. […] This suggests that ML methods would be preferable for prediction models only if a large dataset is available. […] This study showed that ML could achieve great performance in predicting asthma exacerbations.
  • #24 Machine learning for prediction of asthma exacerbations among asthmatic patients: a systematic review and meta-analysis | BMC Pulmonary Medicine | Full Text
    https://bmcpulmmed.biomedcentral.com/articles/10.1186/s12890-023-02570-w
    The overall pooled AUROC (0.8, 95% CI 0.760.83) and DOR (7.02, 95% CI 5.209.47) indicated that ML-based prediction models for asthma exacerbation could achieve good discrimination. […] ML prediction models could forecast patients at high risk of exacerbation from several days to years, helping identify patients needing closer management. […] The sample size is crucial for model performance. […] This suggests that ML methods would be preferable for prediction models only if a large dataset is available. […] This study showed that ML could achieve great performance in predicting asthma exacerbations.
  • #25 JMIR AI – Machine Learning–Based Asthma Attack Prediction Models From Routinely Collected Electronic Health Records: Systematic Scoping Review
    https://ai.jmir.org/2023/1/e46717
    Our review indicates that this research field is still underdeveloped, given the limited body of evidence, heterogeneity of methods, lack of external validation, and suboptimally reported models. There was considerable heterogeneity in the specific definition of asthma outcome and the associated time horizon used by studies that sought to develop asthma attack risk prediction models. Class imbalance was also common across studies, and there was also considerable heterogeneity in how it was handled. Consequently, it was challenging to directly compare the studies. […] The GBDT-based methods were the most reported best-performing method. However, none of the studies was prospectively evaluated or followed any reporting guidelines, and most studies were not externally validated. […] This review highlighted several technical challenges that need to be addressed when developing asthma attack risk prediction algorithms. Further studies are required to develop a robust strategy for dealing with the class imbalance in asthma research. There remains a notable gap in the literature regarding a systematic comparison of the effectiveness of existing methods, particularly in the context of asthma attack prediction.
  • #26 Predicting asthma attacks in primary care: protocol for developing a machine learning-based prediction model | BMJ Open
    https://bmjopen.bmj.com/content/9/7/e028375
    We aim to create a personalised risk assessment tool to assist primary care clinicians in predicting asthma attacks over a period of 1, 4, 12, 26, and 52 weeks, employing machine-learning methodologies such as nave Bayes classifiers, random forests, and support vector machines, as well as ensemble algorithms. […] The model will build on previous research to improve the sensitivity of our event prediction, without unduly compromising the specificity. […] This project will further advance asthma attack risk prediction modelling and will inform on the future direction of routine data linkage in Scotland, which is likely to have additional benefits for other health systems in the UK and internationally.
  • #27 Predicting asthma attacks in primary care: protocol for developing a machine learning-based prediction model | BMJ Open
    https://bmjopen.bmj.com/content/9/7/e028375
    While it might seem intuitive that the patients with most severe daily symptoms exhibit greater risk of severe morbidity and mortality, research suggests that these symptoms may be a suboptimal clinical marker of asthma attack risk. […] Indeed, some people with asthma are more prone to asthma attacks than others, with asthma attack history being the strongest risk factor for future asthma attacks. […] Despite the identification of many risk factors, identifying high-risk individuals has proven a challenging task. […] As such, most prediction models report high specificity (correctly predicting low attack risk to those who did not have attacks), but low sensitivity (correctly predicting high risk in those who did go on to have attacks), which results in less reliable risk prediction for patients at high risk.
  • #28 Predicting asthma attacks in primary care: protocol for developing a machine learning-based prediction model | BMJ Open
    https://bmjopen.bmj.com/content/9/7/e028375
    While it might seem intuitive that the patients with most severe daily symptoms exhibit greater risk of severe morbidity and mortality, research suggests that these symptoms may be a suboptimal clinical marker of asthma attack risk. […] Indeed, some people with asthma are more prone to asthma attacks than others, with asthma attack history being the strongest risk factor for future asthma attacks. […] Despite the identification of many risk factors, identifying high-risk individuals has proven a challenging task. […] As such, most prediction models report high specificity (correctly predicting low attack risk to those who did not have attacks), but low sensitivity (correctly predicting high risk in those who did go on to have attacks), which results in less reliable risk prediction for patients at high risk.
  • #29 JMIR AI – Machine Learning–Based Asthma Attack Prediction Models From Routinely Collected Electronic Health Records: Systematic Scoping Review
    https://ai.jmir.org/2023/1/e46717
    ML model development for asthma attack prediction has been studied in recent years and includes the use of both traditional and DL methods. There is considerable heterogeneity in ML pipelines across existing studies that prohibits meaningful comparison. Our review indicates several key technical challenges that need to be tackled to make progress toward clinical implementation such as class imbalance problem, external validation, model explanation, and adherence to reporting guidelines for model reproducibility.
  • #30 DIGIPREDICT: physiological, behavioural and environmental predictors of asthma attacks—a prospective observational study using digital markers and artificial intelligence—study protocol | BMJ Open Respiratory Research
    https://bmjopenrespres.bmj.com/content/11/1/e002275
    Asthma attacks are a leading cause of morbidity and mortality but are preventable in most if detected and treated promptly. […] The aim of this study DIGIPREDICT is to identify early digital markers of asthma attacks using sensors embedded in smart devices including watches and inhalers, and leverage health and environmental datasets and artificial intelligence, to develop a risk prediction model to provide an early, personalised warning of asthma attacks. […] Identifying these markers will enable early detection and management of attacks and inform the development of a risk prediction model for asthma attacks based on these digital markers. […] The aim of this study is to prospectively collect real-time data on digital biomarkers and other parameters that are associated with asthma attack risk, using digital technologies and wireless sensors, and combine this information with retrospective data from NZ health and environmental datasets to improve prediction of asthma attack risk. […] We hypothesise that there will be clear digital markers that will be associated with asthma attacks, and that their inclusion in a risk prediction model will improve the predictive ability of the models for asthma attacks.
  • #31 DIGIPREDICT: physiological, behavioural and environmental predictors of asthma attacks—a prospective observational study using digital markers and artificial intelligence—study protocol | BMJ Open Respiratory Research
    https://bmjopenrespres.bmj.com/content/11/1/e002275
    Asthma attacks are a leading cause of morbidity and mortality but are preventable in most if detected and treated promptly. […] The aim of this study DIGIPREDICT is to identify early digital markers of asthma attacks using sensors embedded in smart devices including watches and inhalers, and leverage health and environmental datasets and artificial intelligence, to develop a risk prediction model to provide an early, personalised warning of asthma attacks. […] Identifying these markers will enable early detection and management of attacks and inform the development of a risk prediction model for asthma attacks based on these digital markers. […] The aim of this study is to prospectively collect real-time data on digital biomarkers and other parameters that are associated with asthma attack risk, using digital technologies and wireless sensors, and combine this information with retrospective data from NZ health and environmental datasets to improve prediction of asthma attack risk. […] We hypothesise that there will be clear digital markers that will be associated with asthma attacks, and that their inclusion in a risk prediction model will improve the predictive ability of the models for asthma attacks.
  • #32 Adaptive Smart eHealth Framework for Personalized Asthma Attack Prediction and Safe Route Recommendation
    https://www.mdpi.com/2624-6511/6/5/130
    Adaptive Smart eHealth Framework for Personalized Asthma Attack Prediction and Safe Route Recommendation […] Recently, there has been growing interest in using smart eHealth systems to manage asthma. […] The proposed smart eHealth system predicts asthma attacks and uses spatial data to provide a safe route that drives the patient away from any asthma trigger. […] The developed telemonitoring application collected a dataset containing 665 records used to train the prediction models. The testing result demonstrates a remarkable 98% accuracy in predicting asthma attacks with a recall of 96%. […] The proposed system offers numerous advantages over existing asthma management systems. It is proactive, allowing patients to take preventive measures before an asthma attack occurs. […] The proposed system aims to develop a personalized asthma attack prediction model using advanced machine learning techniques. […] The safe route recommendation model of the proposed system is a noteworthy addition to the field of asthma care. […] The proposed system incorporates a crucial feature of system adaptation, which enables it to continuously improve its predictions and recommendations based on new data and user feedback. […] The performance of the asthma attack prediction model is 98% of accuracy and 96.8% recall. […] The performance of the route’s risk level prediction model is 94% of accuracy and 95.2% of recall. […] User feedback revealed that the system is 95% accurate, and 89% of users were satisfied with the safer recommended route compared to their usual route.
  • #33 Adaptive Smart eHealth Framework for Personalized Asthma Attack Prediction and Safe Route Recommendation
    https://www.mdpi.com/2624-6511/6/5/130
    Adaptive Smart eHealth Framework for Personalized Asthma Attack Prediction and Safe Route Recommendation […] Recently, there has been growing interest in using smart eHealth systems to manage asthma. […] The proposed smart eHealth system predicts asthma attacks and uses spatial data to provide a safe route that drives the patient away from any asthma trigger. […] The developed telemonitoring application collected a dataset containing 665 records used to train the prediction models. The testing result demonstrates a remarkable 98% accuracy in predicting asthma attacks with a recall of 96%. […] The proposed system offers numerous advantages over existing asthma management systems. It is proactive, allowing patients to take preventive measures before an asthma attack occurs. […] The proposed system aims to develop a personalized asthma attack prediction model using advanced machine learning techniques. […] The safe route recommendation model of the proposed system is a noteworthy addition to the field of asthma care. […] The proposed system incorporates a crucial feature of system adaptation, which enables it to continuously improve its predictions and recommendations based on new data and user feedback. […] The performance of the asthma attack prediction model is 98% of accuracy and 96.8% recall. […] The performance of the route’s risk level prediction model is 94% of accuracy and 95.2% of recall. […] User feedback revealed that the system is 95% accurate, and 89% of users were satisfied with the safer recommended route compared to their usual route.
  • #34 Adaptive Smart eHealth Framework for Personalized Asthma Attack Prediction and Safe Route Recommendation
    https://www.mdpi.com/2624-6511/6/5/130
    Adaptive Smart eHealth Framework for Personalized Asthma Attack Prediction and Safe Route Recommendation […] Recently, there has been growing interest in using smart eHealth systems to manage asthma. […] The proposed smart eHealth system predicts asthma attacks and uses spatial data to provide a safe route that drives the patient away from any asthma trigger. […] The developed telemonitoring application collected a dataset containing 665 records used to train the prediction models. The testing result demonstrates a remarkable 98% accuracy in predicting asthma attacks with a recall of 96%. […] The proposed system offers numerous advantages over existing asthma management systems. It is proactive, allowing patients to take preventive measures before an asthma attack occurs. […] The proposed system aims to develop a personalized asthma attack prediction model using advanced machine learning techniques. […] The safe route recommendation model of the proposed system is a noteworthy addition to the field of asthma care. […] The proposed system incorporates a crucial feature of system adaptation, which enables it to continuously improve its predictions and recommendations based on new data and user feedback. […] The performance of the asthma attack prediction model is 98% of accuracy and 96.8% recall. […] The performance of the route’s risk level prediction model is 94% of accuracy and 95.2% of recall. […] User feedback revealed that the system is 95% accurate, and 89% of users were satisfied with the safer recommended route compared to their usual route.
  • #35 An Interpretable and Generalizable Machine Learning Model for Predicting Asthma Outcomes: Integrating AutoML and Explainable AI Techniques
    https://www.mdpi.com/2673-4060/6/1/15
    Asthma remains a prevalent chronic condition, impacting millions globally and presenting significant clinical and economic challenges. This study develops a predictive model for asthma outcomes, leveraging automated machine learning (AutoML) and explainable AI (XAI) to balance high predictive accuracy with interpretability. […] Such predictive tools could significantly improve patient outcomes by enabling early intervention and customized management strategies tailored to individual risk profiles. […] Despite advancements in ML and AutoML for healthcare applications, most predictive models for asthma have been validated within isolated datasets, limiting their generalizability. Generalizable models are essential to ensure consistent performance across various populations and clinical settings.
  • #36 An Interpretable and Generalizable Machine Learning Model for Predicting Asthma Outcomes: Integrating AutoML and Explainable AI Techniques
    https://www.mdpi.com/2673-4060/6/1/15
    Asthma remains a prevalent chronic condition, impacting millions globally and presenting significant clinical and economic challenges. This study develops a predictive model for asthma outcomes, leveraging automated machine learning (AutoML) and explainable AI (XAI) to balance high predictive accuracy with interpretability. […] Such predictive tools could significantly improve patient outcomes by enabling early intervention and customized management strategies tailored to individual risk profiles. […] Despite advancements in ML and AutoML for healthcare applications, most predictive models for asthma have been validated within isolated datasets, limiting their generalizability. Generalizable models are essential to ensure consistent performance across various populations and clinical settings.
  • #37 An Interpretable and Generalizable Machine Learning Model for Predicting Asthma Outcomes: Integrating AutoML and Explainable AI Techniques
    https://www.mdpi.com/2673-4060/6/1/15
    Asthma remains a prevalent chronic condition, impacting millions globally and presenting significant clinical and economic challenges. This study develops a predictive model for asthma outcomes, leveraging automated machine learning (AutoML) and explainable AI (XAI) to balance high predictive accuracy with interpretability. […] Such predictive tools could significantly improve patient outcomes by enabling early intervention and customized management strategies tailored to individual risk profiles. […] Despite advancements in ML and AutoML for healthcare applications, most predictive models for asthma have been validated within isolated datasets, limiting their generalizability. Generalizable models are essential to ensure consistent performance across various populations and clinical settings.
  • #38 Early Prediction of Asthma
    https://pmc.ncbi.nlm.nih.gov/articles/PMC10455492/
    The clinical manifestations of asthma in children are highly variable, are associated with different molecular and cellular mechanisms, and are characterized by common symptoms that may diversify in frequency and intensity throughout life. […] It is essential to promptly identify patients at high risk of developing asthma by using different prediction models. […] This review also highlights the epigenetic factors that are involved, such as DNA methylation and asthma risk, microRNA expression, and histone modification. […] The early identification of wheezing in children younger than five years old can provide valuable information to parents and medical professionals and aid in the early stratification and close monitoring of patients at risk of asthma. […] Prediction models for childhood asthma are helpful for identifying likely future asthma patients from high-risk groups; children in preschool who develop symptoms could benefit from early diagnosis and intervention.
  • #39 Early Prediction of Asthma
    https://pmc.ncbi.nlm.nih.gov/articles/PMC10455492/
    The clinical manifestations of asthma in children are highly variable, are associated with different molecular and cellular mechanisms, and are characterized by common symptoms that may diversify in frequency and intensity throughout life. […] It is essential to promptly identify patients at high risk of developing asthma by using different prediction models. […] This review also highlights the epigenetic factors that are involved, such as DNA methylation and asthma risk, microRNA expression, and histone modification. […] The early identification of wheezing in children younger than five years old can provide valuable information to parents and medical professionals and aid in the early stratification and close monitoring of patients at risk of asthma. […] Prediction models for childhood asthma are helpful for identifying likely future asthma patients from high-risk groups; children in preschool who develop symptoms could benefit from early diagnosis and intervention.
  • #40 Early Prediction of Asthma
    https://pmc.ncbi.nlm.nih.gov/articles/PMC10455492/
    According to a study of a population that was part of FinEsS (Finland-Estonia-Sweden), the median age for the diagnosis of allergic asthma was 19 years, and that for non-allergic asthma was 35 years. […] The incidence of allergic asthma was high in the 0-9 age group (1.8/1000/year) and lower in the 50-59 age group (0.6/1000/year). […] The allergic phenotype associated with asthma exhibits a diverse range of characteristics that are influenced by a complex interplay of environmental, genetic, and psychosocial factors. […] Familial atopy is consistently recognized as a significant predictor of asthma from childhood to adulthood, with the children of allergic parents manifesting rates of asthma that are two to three times higher. […] Prediction models for childhood asthma have been proven to be functional in recognizing future asthmatics in high-risk groups of patients through their use in the preschool period, which is a crucial period for immune development and lung growth. […] Current knowledge of epigenetics, the use of biomarkers, and different types of algorithms in the prediction of asthma in children provide an opportunity to improve the accuracy of these diagnostic tools.
  • #41 Early Prediction of Asthma
    https://pmc.ncbi.nlm.nih.gov/articles/PMC10455492/
    According to a study of a population that was part of FinEsS (Finland-Estonia-Sweden), the median age for the diagnosis of allergic asthma was 19 years, and that for non-allergic asthma was 35 years. […] The incidence of allergic asthma was high in the 0-9 age group (1.8/1000/year) and lower in the 50-59 age group (0.6/1000/year). […] The allergic phenotype associated with asthma exhibits a diverse range of characteristics that are influenced by a complex interplay of environmental, genetic, and psychosocial factors. […] Familial atopy is consistently recognized as a significant predictor of asthma from childhood to adulthood, with the children of allergic parents manifesting rates of asthma that are two to three times higher. […] Prediction models for childhood asthma have been proven to be functional in recognizing future asthmatics in high-risk groups of patients through their use in the preschool period, which is a crucial period for immune development and lung growth. […] Current knowledge of epigenetics, the use of biomarkers, and different types of algorithms in the prediction of asthma in children provide an opportunity to improve the accuracy of these diagnostic tools.
  • #42 Early Prediction of Asthma
    https://pmc.ncbi.nlm.nih.gov/articles/PMC10455492/
    According to a study of a population that was part of FinEsS (Finland-Estonia-Sweden), the median age for the diagnosis of allergic asthma was 19 years, and that for non-allergic asthma was 35 years. […] The incidence of allergic asthma was high in the 0-9 age group (1.8/1000/year) and lower in the 50-59 age group (0.6/1000/year). […] The allergic phenotype associated with asthma exhibits a diverse range of characteristics that are influenced by a complex interplay of environmental, genetic, and psychosocial factors. […] Familial atopy is consistently recognized as a significant predictor of asthma from childhood to adulthood, with the children of allergic parents manifesting rates of asthma that are two to three times higher. […] Prediction models for childhood asthma have been proven to be functional in recognizing future asthmatics in high-risk groups of patients through their use in the preschool period, which is a crucial period for immune development and lung growth. […] Current knowledge of epigenetics, the use of biomarkers, and different types of algorithms in the prediction of asthma in children provide an opportunity to improve the accuracy of these diagnostic tools.
  • #43 Early Prediction of Asthma
    https://pmc.ncbi.nlm.nih.gov/articles/PMC10455492/
    According to a study of a population that was part of FinEsS (Finland-Estonia-Sweden), the median age for the diagnosis of allergic asthma was 19 years, and that for non-allergic asthma was 35 years. […] The incidence of allergic asthma was high in the 0-9 age group (1.8/1000/year) and lower in the 50-59 age group (0.6/1000/year). […] The allergic phenotype associated with asthma exhibits a diverse range of characteristics that are influenced by a complex interplay of environmental, genetic, and psychosocial factors. […] Familial atopy is consistently recognized as a significant predictor of asthma from childhood to adulthood, with the children of allergic parents manifesting rates of asthma that are two to three times higher. […] Prediction models for childhood asthma have been proven to be functional in recognizing future asthmatics in high-risk groups of patients through their use in the preschool period, which is a crucial period for immune development and lung growth. […] Current knowledge of epigenetics, the use of biomarkers, and different types of algorithms in the prediction of asthma in children provide an opportunity to improve the accuracy of these diagnostic tools.
  • #44 Early Prediction of Asthma
    https://pmc.ncbi.nlm.nih.gov/articles/PMC10455492/
    According to a study of a population that was part of FinEsS (Finland-Estonia-Sweden), the median age for the diagnosis of allergic asthma was 19 years, and that for non-allergic asthma was 35 years. […] The incidence of allergic asthma was high in the 0-9 age group (1.8/1000/year) and lower in the 50-59 age group (0.6/1000/year). […] The allergic phenotype associated with asthma exhibits a diverse range of characteristics that are influenced by a complex interplay of environmental, genetic, and psychosocial factors. […] Familial atopy is consistently recognized as a significant predictor of asthma from childhood to adulthood, with the children of allergic parents manifesting rates of asthma that are two to three times higher. […] Prediction models for childhood asthma have been proven to be functional in recognizing future asthmatics in high-risk groups of patients through their use in the preschool period, which is a crucial period for immune development and lung growth. […] Current knowledge of epigenetics, the use of biomarkers, and different types of algorithms in the prediction of asthma in children provide an opportunity to improve the accuracy of these diagnostic tools.
  • #45 Predicting exacerbation in young children with intermittent asthm
    https://www.openaccessjournals.com/articles/predicting-exacerbation-in-young-children-with-intermittent-asthma-12480.html
    Levels of FEV1, FEV1/VC and PD20FEV1 below cut off levels are highly predictive for a consecutive AE. […] The presence of two or three risk factors was associated with a significantly higher risk of AE. […] With 2 out of these 3 risk factors, the OR for an AE increased to 5.25 and reached 11.3 with 3 risk factors. […] Our results demonstrate that levels of FEV1, FEV1/FVC and PD20FEV1 below the cut-off levels are highly predictive for a consecutive AE.
  • #46 Predicting exacerbation in young children with intermittent asthm
    https://www.openaccessjournals.com/articles/predicting-exacerbation-in-young-children-with-intermittent-asthma-12480.html
    Levels of FEV1, FEV1/VC and PD20FEV1 below cut off levels are highly predictive for a consecutive AE. […] The presence of two or three risk factors was associated with a significantly higher risk of AE. […] With 2 out of these 3 risk factors, the OR for an AE increased to 5.25 and reached 11.3 with 3 risk factors. […] Our results demonstrate that levels of FEV1, FEV1/FVC and PD20FEV1 below the cut-off levels are highly predictive for a consecutive AE.
  • #47 Predicting exacerbation in young children with intermittent asthm
    https://www.openaccessjournals.com/articles/predicting-exacerbation-in-young-children-with-intermittent-asthma-12480.html
    Levels of FEV1, FEV1/VC and PD20FEV1 below cut off levels are highly predictive for a consecutive AE. […] The presence of two or three risk factors was associated with a significantly higher risk of AE. […] With 2 out of these 3 risk factors, the OR for an AE increased to 5.25 and reached 11.3 with 3 risk factors. […] Our results demonstrate that levels of FEV1, FEV1/FVC and PD20FEV1 below the cut-off levels are highly predictive for a consecutive AE.
  • #48 Seasonal variation of pediatric asthma exacerbations and its association with asthma phenotypes | Pediatric Research
    https://www.nature.com/articles/s41390-025-04073-2
    We further identified associations between asthma blood inflammatory phenotypes and an increase in pediatric asthma exacerbations before the COVID-19 pandemic during winter and autumn seasons. […] In conclusion, our study highlights that the HBE/LBN phenotype was associated with increased asthma exacerbations among hospitalized pediatric asthma patients during the winter and autumn seasons.
  • #49 Seasonal variation of pediatric asthma exacerbations and its association with asthma phenotypes | Pediatric Research
    https://www.nature.com/articles/s41390-025-04073-2
    We further identified associations between asthma blood inflammatory phenotypes and an increase in pediatric asthma exacerbations before the COVID-19 pandemic during winter and autumn seasons. […] In conclusion, our study highlights that the HBE/LBN phenotype was associated with increased asthma exacerbations among hospitalized pediatric asthma patients during the winter and autumn seasons.
  • #50 Applying UK real-world primary care data to predict asthma attacks in 3776 well-characterised children: a retrospective cohort study | npj Primary Care Respiratory Medicine
    https://www.nature.com/articles/s41533-018-0095-5
    Current understanding of risk factors for asthma attacks in children is based on studies of small but well-characterised populations or pharmaco-epidemiology studies of large but poorly characterised populations. […] In this large population, several factors were associated with a future asthma attack, but a past history of attacks was most strongly associated with future attacks. […] Our results indicate that, of all the outcomes collected in this large study, a past asthma attack (and especially two attacks) is likely to be the best method to identify children who might benefit from a stratified intervention aimed at reducing their risk for future asthma attacks. […] Factors other than a past history of attacks which were associated with increased risk for future asthma attacks included blood eosinophilia, reduced PEF, lower respiratory tract infection and younger age, and although these associations were highly significant, they were weakly related to risk for asthma attacks and therefore not likely to be particularly helpful in risk analysis. […] In summary, we find that a past history of asthma attacks is the best predictor of future attacks, and that blood eosinophilia and reduced PEF do not add substantially to predicting attacks.
  • #51 Machine learning for prediction of asthma exacerbations among asthmatic patients: a systematic review and meta-analysis | BMC Pulmonary Medicine | Full Text
    https://bmcpulmmed.biomedcentral.com/articles/10.1186/s12890-023-02570-w
    Asthma exacerbations reduce the patients quality of life and are also responsible for significant disease burdens and economic costs. […] This systematic review and meta-analysis aimed to identify the prediction performance of ML-based prediction models for asthma exacerbations and address the uncertainty of whether modern ML methods could become an alternative option to predict asthma exacerbations. […] The overall pooled area under the curve of the receiver operating characteristics (AUROC) of 11 studies (23 prediction models) was 0.80 (95% CI 0.770.83). […] This study identified that ML was the potential tool to achieve great performance in predicting asthma exacerbations. […] Future studies should focus on improving the generalization ability and practicability, thus driving the application of these models in clinical practice.
  • #52 Acute severe asthma in emergency department: clinical characteristics, risk factors, and predictors for poor outcome | The Egyptian Journal of Bronchology | Full Text
    https://ejb.springeropen.com/articles/10.1186/s43168-022-00160-8
    Severe asthma exacerbations can be predicted by old age, previous history of mechanical ventilation, obstructive sleep apnea, overuse of SABA, uncontrolled asthma, moderate to severe depression, eosinophilia, SO2 90%, and low peak expiratory flow rates. […] The most important predictors for severe exacerbations were SO2 90% at baseline (OR = 4.56; 95% CI = 3.457.56; P0.001), PEFR after 1 h (OR= 3.34; 95%CI = 1.904.90; P0.001), and uncontrolled asthma (OR= 3.33; 95%CI = 2.505.05; P0.001). […] Based on the current study, the predictors of hospitalization were old age (OR= 1.11; 95%CI= 1.092.11; P0.001), uncontrolled asthma (OR= 2.34; 95%CI= 2.014.40; P0.001), PEFR after 1 h (OR= 4.44; 95%CI= 3.247.68; P0.001), and SO2 90% at baseline (OR= 5.67; 95%CI= 3.988.50; P0.001). […] The most accurate independent predictor for severe asthma exacerbation is the PEFR value after 1 h of treatment.
  • #53 Seasonal variation of pediatric asthma exacerbations and its association with asthma phenotypes | Pediatric Research
    https://www.nature.com/articles/s41390-025-04073-2
    HBE/LBN phenotype had a higher risk of asthma exacerbations among hospitalized pediatric asthma patients in the winter and autumn, while LBE/LBN phenotype had a lower risk in the winter, spring, and summer. […] Blood eosinophils and neutrophils have been indicated to have a potential influence on pediatric asthma development and severity. […] HBE/LBN phenotype was associated with increased asthma exacerbations among hospitalized pediatric asthma patients during winter and autumn. […] Eosinophil and neutrophil predominance exhibited a higher influence on pediatric asthma exacerbations. […] We observed that asthma phenotypes were associated with an increase in asthma exacerbations in hospitalized pediatric asthma patients. […] Our study further identified that asthma blood inflammatory phenotypes were associated with an increase in asthma exacerbations of hospitalized pediatric asthma patients during winter and autumn.
  • #54 DIGIPREDICT: physiological, behavioural and environmental predictors of asthma attacks—a prospective observational study using digital markers and artificial intelligence—study protocol | BMJ Open Respiratory Research
    https://bmjopenrespres.bmj.com/content/11/1/e002275
    Asthma attacks are a leading cause of morbidity and mortality but are preventable in most if detected and treated promptly. […] The aim of this study DIGIPREDICT is to identify early digital markers of asthma attacks using sensors embedded in smart devices including watches and inhalers, and leverage health and environmental datasets and artificial intelligence, to develop a risk prediction model to provide an early, personalised warning of asthma attacks. […] Identifying these markers will enable early detection and management of attacks and inform the development of a risk prediction model for asthma attacks based on these digital markers. […] The aim of this study is to prospectively collect real-time data on digital biomarkers and other parameters that are associated with asthma attack risk, using digital technologies and wireless sensors, and combine this information with retrospective data from NZ health and environmental datasets to improve prediction of asthma attack risk. […] We hypothesise that there will be clear digital markers that will be associated with asthma attacks, and that their inclusion in a risk prediction model will improve the predictive ability of the models for asthma attacks.
  • #55 Adaptive Smart eHealth Framework for Personalized Asthma Attack Prediction and Safe Route Recommendation
    https://www.mdpi.com/2624-6511/6/5/130
    Adaptive Smart eHealth Framework for Personalized Asthma Attack Prediction and Safe Route Recommendation […] Recently, there has been growing interest in using smart eHealth systems to manage asthma. […] The proposed smart eHealth system predicts asthma attacks and uses spatial data to provide a safe route that drives the patient away from any asthma trigger. […] The developed telemonitoring application collected a dataset containing 665 records used to train the prediction models. The testing result demonstrates a remarkable 98% accuracy in predicting asthma attacks with a recall of 96%. […] The proposed system offers numerous advantages over existing asthma management systems. It is proactive, allowing patients to take preventive measures before an asthma attack occurs. […] The proposed system aims to develop a personalized asthma attack prediction model using advanced machine learning techniques. […] The safe route recommendation model of the proposed system is a noteworthy addition to the field of asthma care. […] The proposed system incorporates a crucial feature of system adaptation, which enables it to continuously improve its predictions and recommendations based on new data and user feedback. […] The performance of the asthma attack prediction model is 98% of accuracy and 96.8% recall. […] The performance of the route’s risk level prediction model is 94% of accuracy and 95.2% of recall. […] User feedback revealed that the system is 95% accurate, and 89% of users were satisfied with the safer recommended route compared to their usual route.
  • #56 An Interpretable and Generalizable Machine Learning Model for Predicting Asthma Outcomes: Integrating AutoML and Explainable AI Techniques
    https://www.mdpi.com/2673-4060/6/1/15
    Asthma remains a prevalent chronic condition, impacting millions globally and presenting significant clinical and economic challenges. This study develops a predictive model for asthma outcomes, leveraging automated machine learning (AutoML) and explainable AI (XAI) to balance high predictive accuracy with interpretability. […] Such predictive tools could significantly improve patient outcomes by enabling early intervention and customized management strategies tailored to individual risk profiles. […] Despite advancements in ML and AutoML for healthcare applications, most predictive models for asthma have been validated within isolated datasets, limiting their generalizability. Generalizable models are essential to ensure consistent performance across various populations and clinical settings.
  • #57 Early Prediction of Asthma
    https://pmc.ncbi.nlm.nih.gov/articles/PMC10455492/
    The clinical manifestations of asthma in children are highly variable, are associated with different molecular and cellular mechanisms, and are characterized by common symptoms that may diversify in frequency and intensity throughout life. […] It is essential to promptly identify patients at high risk of developing asthma by using different prediction models. […] This review also highlights the epigenetic factors that are involved, such as DNA methylation and asthma risk, microRNA expression, and histone modification. […] The early identification of wheezing in children younger than five years old can provide valuable information to parents and medical professionals and aid in the early stratification and close monitoring of patients at risk of asthma. […] Prediction models for childhood asthma are helpful for identifying likely future asthma patients from high-risk groups; children in preschool who develop symptoms could benefit from early diagnosis and intervention.
  • #58 JMIR AI – Machine Learning–Based Asthma Attack Prediction Models From Routinely Collected Electronic Health Records: Systematic Scoping Review
    https://ai.jmir.org/2023/1/e46717
    ML model development for asthma attack prediction has been studied in recent years and includes the use of both traditional and DL methods. There is considerable heterogeneity in ML pipelines across existing studies that prohibits meaningful comparison. Our review indicates several key technical challenges that need to be tackled to make progress toward clinical implementation such as class imbalance problem, external validation, model explanation, and adherence to reporting guidelines for model reproducibility.
  • #59 Primary Care Asthma Attack Prediction Models for Adults | JAA
    https://www.dovepress.com/primary-care-asthma-attack-prediction-models-for-adults-a-systematic-r-peer-reviewed-fulltext-article-JAA
    Prognostic models hold great potential for predicting asthma exacerbations, providing opportunities for early intervention, and are a popular area of current research. […] This systematic review aimed to identify novel predictive models of asthma attacks in adults and compare differences in construction related to populations, outcome definitions, prediction time horizons, algorithms, validation, and performance estimation. […] The outcome of the prediction model was the onset of an asthma attack, and studies that only reported other asthma-related events (such as post-asthma attack hospital discharge) or statuses (such as uncontrolled asthma or asthma severity) were excluded. […] The predictive performance is heavily influenced by the study design, including the population, the outcome definition, the algorithm, and the model validation procedures. Identifying the most clinically meaningful model characteristics is necessary to enable a best model to be identified and highlight routes for future development. This will boost likelihood of successful translation, adoption, and implementation at scale of clinical prediction models, and to bring benefits to patients.

Zobacz więcej

infekcja dolnych dróg oddechowych iloraz szans szczytowy przepływ wydechowy cyfrowy biomarker obturacyjny bezdech senny maszyna wektorów nośnych metylacja DNA modyfikacja histonów niekontrolowana astma las losowy świszczący oddech astma alergiczna sztuczna inteligencja eozynofilia saturacja tlenu ciężkie zaostrzenie astmy wzrost płuc astma niealergiczna fenotyp astmy marker zapalny walidacja zewnętrzna ekspresja mikroRNA zaostrzenie astmy uczenie maszynowe krótko działający beta-agonista model predykcyjny atak astmy