Covid-19 – Rokowania, prognozy i postęp choroby

Covid-19
Rokowania, prognozy i postęp choroby

Prognozowanie przebiegu klinicznego COVID-19 jest kluczowe dla optymalizacji leczenia i alokacji zasobów medycznych. W literaturze opisano ponad 600 modeli prognostycznych, z których większość skupia się na przewidywaniu śmiertelności i ciężkiego przebiegu choroby. Najważniejsze predyktory to podeszły wiek, płeć, obniżone wysycenie tlenem, podwyższone poziomy CRP (do 69,10 mg/L), BUN, D-dimerów (DoM 1,29 mg/L), IL-6, ferrytyny, LDH (DoM 189,49 U/L) i troponiny sercowej I (DoM 21,88 pg/mL). Modele oparte na uczeniu maszynowym, takie jak połączone algorytmy SVM, Gradient Boosted Decision Tree i sieci neuronowe, osiągają wysokie wartości AUC (do 0,99) w przewidywaniu ciężkiego przebiegu i śmiertelności. Wśród skal klinicznych, 4C mortality score i COVID-IRS wykazują dobrą dokładność (AUC odpowiednio 0,83 i 0,78), a nowa skala COVID-COMBI przewyższa je w predykcji ciężkiego przebiegu (AUC 0,79).

Wprowadzenie do prognozy COVID-19

Trafne przewidywanie przebiegu klinicznego u pacjentów z COVID-19 stanowi kluczowe wyzwanie dla współczesnej medycyny, zarówno po ostrej infekcji, jak i podczas długoterminowej obserwacji. Wczesna identyfikacja pacjentów zagrożonych ciężkim przebiegiem choroby ma krytyczne znaczenie dla szybkiego wdrożenia odpowiedniego leczenia i efektywnej alokacji zasobów medycznych.¹² Modele prognostyczne mogą wspierać personel medyczny w podejmowaniu decyzji klinicznych i umożliwiać stratyfikację ryzyka pacjentów z COVID-19, co potencjalnie może przełożyć się na zmniejszenie wskaźników śmiertelności.³

Typy modeli prognostycznych w COVID-19

Modele prognostyczne dla COVID-19 zostały wprowadzone do literatury naukowej w bezprecedensowym tempie i w dużej liczbie, aby wspierać podejmowanie decyzji medycznych.⁴ Zidentyfikowano ponad 600 modeli prognostycznych, z czego większość (593) koncentruje się na przewidywaniu różnych wyników u pacjentów z potwierdzonym zakażeniem, a pozostałe (13) na identyfikacji osób z populacji ogólnej narażonych na zwiększone ryzyko infekcji, hospitalizacji lub zgonu.⁵⁶ Zgłaszane wartości indeksu C wahały się między 0,77 a 0,93 w badaniach rozwojowych z niskim ryzykiem błędu oraz między 0,56 a 0,78 w zewnętrznych walidacjach z niskim ryzykiem błędu.⁷

Modele prognostyczne śmiertelności

Najczęstszym przewidywanym punktem końcowym w modelach prognostycznych COVID-19 była śmiertelność, oceniana w 152 (48,4%) badaniach.⁸ Wśród badań, które określały i raportowały czas predykcji dla śmiertelności, występowały różnice w raportowaniu, które można podzielić na modele z ustalonym przedziałem czasowym oraz modele czasu do wystąpienia zdarzenia.⁹ Te drugie biorą pod uwagę nie tylko wystąpienie zdarzenia, ale także czas, w którym ono następuje.¹⁰

Najczęściej występującymi predyktorami śmiertelności w omawianych 152 badaniach były: podeszły wiek, płeć, obniżone wysycenie tlenem, podwyższone poziomy białka C-reaktywnego (CRP), azotu mocznikowego we krwi (BUN), temperatura ciała, liczba chorób współistniejących, zaburzenia świadomości, liczba białych krwinek, liczba limfocytów, poziom D-dimerów, liczba płytek krwi oraz częstość tętna.¹¹

Modele prognostyczne ciężkiego przebiegu

Drugim najczęściej przewidywanym punktem końcowym był ciężki przebieg lub stan krytyczny COVID-19, raportowany w 66 (21,0%) badaniach.¹² Najczęściej występującymi predyktorami ciężkiego przebiegu były: podeszły wiek, płeć, temperatura ciała, liczba chorób współistniejących (choroby sercowo-naczyniowe, nadciśnienie, cukrzyca), obniżone wysycenie tlenem, podwyższone poziomy CRP, azotu mocznikowego we krwi, temperatura ciała, skurczowe ciśnienie krwi, stosunek neutrofili do limfocytów (NLR), liczba białych krwinek, liczba limfocytów oraz częstość tętna.¹³

Najlepszą zgłoszoną wydajność predykcyjną dla ciężkiego przebiegu miał połączony model uczenia maszynowego (Support Vector Machine, Gradient Boosted Decision Tree i Neural Network) oparty na prokalcytoninie, liczbie komórek [T+B+NK], IL-6, CRP, receptorze IL-2, limfocytach T-pomocniczych/T-supresorowych jako predyktorach stanu krytycznego, z wartością AUC 0,99.¹⁴

Kluczowe biomarkery prognostyczne w COVID-19

Badania wykazały, że CRP, D-dimery, ferrytyna i IL-6 były istotnie skorelowane z prognozą u pacjentów z ciężkim COVID-19.¹⁵ Model oparty na losowym lesie może przewidzieć z 97% dokładnością prawdopodobieństwo pogorszenia stanu pacjentów z COVID-19 do stanu ciężkiego, przy czym najważniejszymi wskaźnikami były interleukina-6 (IL-6), ferrytyna i D-dimery.¹⁶

W kompleksowej metaanalizie wykazano, że wiek, choroby naczyniowo-mózgowe, CRP, dehydrogenaza mleczanowa (LDH) i troponina sercowa I (cTnI) są najważniejszymi czynnikami ryzyka prognozującymi ciężki przebieg COVID-19.¹⁷ Różnica median (DoM) między osobami, które zmarły, a tymi, które przeżyły, wynosiła 13,15 lat (95% przedział ufności (CI) 11,37 do 14,94) dla wieku, 69,10 mg/L (CI 50,43 do 87,77) dla CRP, 189,49 U/L (CI 155,00 do 223,98) dla LDH, 21,88 pg/mL (CI 9,78 do 33,99) dla cTnI i 1,29 mg/L (CI 0,9 do 1,69) dla D-dimerów.¹⁸

Kombinowane wskaźniki zapalne

Szczególnie interesujące są kombinowane wskaźniki oparte na biomarkerach zapalnych, które mogą poprawić prognozowanie wyników leczenia COVID-19. Badanie przeprowadzone w europejskim ośrodku trzeciorzędowym wykazało, że kombinowane wyniki CRP i innych biomarkerów zapalnych, oparte na wartościach przy przyjęciu, są obiecującymi predyktorami konieczności wspomagania oddechowego przy użyciu kaniuli nosowej o wysokim przepływie (HFNC) oraz śmiertelności u pacjentów cierpiących na ciężkie zakażenie COVID-19.¹⁹

Dwa punkty w skali C-CRP #1 i 2 punkty w skali C-CRP #3 przewidywały konieczność stosowania HFNC z prawdopodobieństwem wynoszącym odpowiednio 0,625 (p = 0,005) i 0,561 (p < 0,001). Dla śmiertelności optymalny punkt odcięcia dla CRP wynosił 1,11, a dla CLR 3,2*10^33. Dwa punkty w skali C-CRP #3* z dokładnością 0,922 przewidywały śmiertelność (p = 0,0038) w ciężkim COVID-19.²⁰

Modele uczenia maszynowego w prognozowaniu COVID-19

Modele uczenia maszynowego (ML) zyskują coraz większe znaczenie w prognozowaniu przebiegu COVID-19. Łączenie danych tabelarycznych i tekstowych może prowadzić do lepszej predykcji negatywnych wyników u pacjentów z COVID-19.²¹ Głównym odkryciem jednego z badań było to, że połączenie zmiennych tabelarycznych i tekstowych prowadziło do lepszego przewidywania 30-dniowej śmiertelności w porównaniu z modelami wykorzystującymi tylko zmienne tabelaryczne.²²

W przypadku przyjęcia na oddział intensywnej terapii (OIT), model łączony miał wyższą precyzję, swoistość, wynik F1 i wartości MCC.²³ Wyniki tego badania sugerują, że połączona analiza danych tabelarycznych i tekstowych była skuteczna w przewidywaniu 30-dniowej śmiertelności u pacjentów z zakażeniem SARS-CoV-2 zgłaszających się na oddział ratunkowy oraz w przewidywaniu ich przyjęcia na OIT.²⁴

Modele oparte na sieciach neuronowych

Sztuczne sieci neuronowe (ANN) wykazują również obiecujące wyniki w prognozowaniu COVID-19. W jednym z badań specyficzna dla pacjenta śmiertelność była przewidywana z dokładnością 86,25%, z czułością 87,50% (95% CI 61,65%-98,45%) i swoistością 85,94% (95% CI 74,98%-93,36%). Pozytywna wartość predykcyjna wynosiła 60,87% (95% CI 45,23%-74,56%), a negatywna wartość predykcyjna 96,49% (95% CI 88,23%-99,02%). Obszar pod krzywą charakterystyki operacyjnej odbiornika (ROC) wynosił 90,12%.²⁵

Zaburzenia świadomości, duszność i podeszły wiek okazały się najbardziej istotnymi ogólnymi cechami w przewidywaniu śmiertelności. Umiarkowanymi predyktorami śmiertelności były zapaść, płeć męska, nowy kaszel i wcześniejsza patologia układu oddechowego.²⁶ Zaletą tego modelu jest to, że wszystkie dane demograficzne, choroby współistniejące, styl życia i objawy mogą być zebrane podczas pierwszego spotkania z lekarzem, dzięki czemu wczesne przewidywanie wyniku może być dokonane po przyjęciu do szpitala.²⁷

Boosted Random Forest

Innym obiecującym podejściem jest zastosowanie algorytmu Boosted Random Forest. Zaproponowano dopracowany model Random Forest wzmocniony algorytmem AdaBoost, który wykorzystuje dane geograficzne, podróżnicze, zdrowotne i demograficzne pacjentów z COVID-19 do przewidywania ciężkości przypadku i możliwego wyniku – wyzdrowienia lub śmierci.²⁸

Modele hybrydowe i strategie zastępcze

W początkowej fazie pandemii, gdy dane indywidualne pacjentów z COVID-19 nie były jeszcze dostępne, istniała potrzeba opracowania predyktorów ryzyka wspierających decyzje dotyczące profilaktyki i leczenia.²⁹ Opracowano strategię hybrydową łączącą rozwój bazowego predyktora ryzyka ciężkiego zakażenia układu oddechowego i metodę przetwarzania wtórnego do kalibracji przewidywań na podstawie zgłaszanych wskaźników śmiertelności dla COVID-19.³⁰

Po zgromadzeniu kohorty pacjentów z COVID-19, przewidywania tego modelu zostały zwalidowane jako mające dobrą dyskryminację (pole pod krzywą ROC wynoszące 0,943) i kalibrację (znacznie lepszą w porównaniu z bazowym predyktorem).³¹ Na zbiorze walidacyjnym składającym się z 4179 pacjentów z COVID-19 (wskaźnik śmiertelności 3,4%) model ten wykazał wysoką zdolność dyskryminacyjną z AUROC wynoszącym 0,943.³²

Walidowane modele i skale ryzyka

Dostępnych jest kilka systemów oceny ryzyka dla COVID-19. W dużej, heterogenicznej kohorcie szwajcarskiej oceniono wydajność skal takich jak National Early Warning Score (NEWS), CURB-65, 4C mortality score (4C), Spanish Society of Infectious Diseases and Clinical Microbiology score (COVID-SEIMC) i COVID Intubation Risk Score (COVID-IRS) u pacjentów hospitalizowanych z powodu COVID-19 w latach 2020 i 2021.³³

Wśród uznanych skal, 4C miała najlepszą dokładność w przewidywaniu ciężkiego przebiegu (AUC 0,76), a następnie COVID-IRS (AUC 0,72). Nowa skala „COVID-COMBI”, łącząca parametry z dwóch najlepszych skal, wykazała znacznie wyższą dokładność niż wszystkie uznane skale (AUC 0,79, p = 0,001). W przypadku przewidywania śmiertelności szpitalnej, 4C wykazał najlepszą wydajność (AUC 0,83), a w przypadku wentylacji mechanicznej najlepiej sprawdził się COVID-IRS (AUC 0,78).³⁴

Badanie to wykazało, że zarówno skala śmiertelności 4C, jak i skala COVID-IRS mają solidną dokładność predykcyjną dla ciężkiego przebiegu u pacjentów hospitalizowanych z powodu COVID-19. Co więcej, nowo opracowana skala COVID-COMBI przewyższyła oba te narzędzia w przewidywaniu ciężkiego przebiegu choroby, co podkreśla jej potencjalną użyteczność w praktyce klinicznej.³⁵

Ograniczenia obecnych modeli prognostycznych

Pomimo dużej liczby dostępnych modeli prognostycznych dla COVID-19, większość z nich była słabo raportowana i obarczona wysokim ryzykiem błędu, co sprawia, że ich zgłaszane wydajności predykcyjne są prawdopodobnie zawyżone.³⁶ Modele z niskim ryzykiem błędu powinny być walidowane przed wdrożeniem klinicznym, najlepiej poprzez wysiłki współpracy, aby umożliwić również zbadanie heterogeniczności ich wydajności w różnych populacjach i warunkach.³⁷

Ogólnie ocena ryzyka błędu (RoB) została oceniona jako wysoka lub niejasna w 312 badaniach.³⁸ Można to wytłumaczyć niedociągnięciami takimi jak słaba jakość metodologiczna, mała wielkość próby, niewłaściwe postępowanie z brakującymi danymi, niepowodzenie w radzeniu sobie z przeuczeniem, definicje COVID-19 oparte na cechach klinicznych, a nie na wyniku laboratoryjnego testu diagnostycznego dla SARS-CoV-2 oraz heterogeniczność punktów końcowych w badaniach.³⁹

Wyzwania w dokładnym prognozowaniu

Warto zauważyć, że żaden model nie może dokładnie przewidzieć wyniku, w tym modele danych dla COVID-19.⁴⁰ Wyniki z dowolnego/wszystkich tych modeli mogą znacznie się różnić przy niewielkiej zmianie założeń, które zasadniczo są założeniami wejściowymi podanymi do modelu dla pożądanego wyniku.⁴¹

Obecne dowody dotyczące modeli prognostycznych COVID-19 są niespójne, a możliwość zastosowania klinicznego pozostaje kontrowersyjna.⁴² Przyszłe duże, wieloośrodkowe i dobrze zaprojektowane badania prospektywne są potrzebne do wyjaśnienia pozostałych niepewności i opracowania modeli predykcyjnych dla COVID-19 o użyteczności klinicznej, które mogą być stosowane w różnych populacjach.⁴³

Wpływ pandemii na wyniki leczenia

Pandemia była związana z gorszymi wynikami klinicznymi dla pacjentów zdiagnozowanych w okresie obostrzeń (Grupa 1) w porównaniu do tych zdiagnozowanych po tym okresie (Grupa 2).⁴⁴ Pacjenci w Grupie 1 mieli gorsze wyniki wzrokowe po 12 miesiącach i przy ostatniej wizycie kontrolnej w porównaniu do Grupy 2, z niższym odsetkiem osiągającym ostrość wzroku ≥20/40 i wyższym odsetkiem z ostrością wzroku ≤20/70.⁴⁵

Te ustalenia podkreślają znaczenie opracowania adaptowalnych strategii utrzymania skutecznej opieki nad pacjentami w obliczu zewnętrznych wyzwań stawianych przez globalną pandemię.⁴⁶

Przyszłość prognozowania COVID-19

Przyszłe badania będą konieczne do poprawy dokładności algorytmów i potwierdzenia ich możliwości uogólnienia poprzez zewnętrzne kohorty walidacyjne, szczególnie w zmieniającej się epidemiologii pandemii.⁴⁷

W kontekście ewoluującej pandemii bez ustalonego systemu punktacji prognostycznej, podejścia oparte na głębokim uczeniu mogą być wykorzystane do szybkiego opracowania empirycznych modeli prognostycznych. Modele te mają nieodłączną zaletę stawania się coraz bardziej dokładnymi i reprezentatywnymi wraz ze wzrostem wielkości zbiorów danych.⁴⁸

Wnioski

Dokładne prognozowanie przebiegu klinicznego u pacjentów z COVID-19 pozostaje wyzwaniem, ale modele oparte na uczeniu maszynowym, biomarkerach i danych klinicznych oferują obiecujące narzędzia do stratyfikacji ryzyka. Identyfikacja kluczowych biomarkerów, takich jak IL-6, ferrytyna, D-dimery, CRP, LDH i cTnI, może pomóc w przewidywaniu ciężkiego przebiegu choroby i śmiertelności. Istnieje potrzeba dalszych badań w celu walidacji i doskonalenia tych modeli, szczególnie w różnych populacjach i warunkach klinicznych, aby zapewnić ich skuteczność i użyteczność w praktyce klinicznej.

Pomimo ograniczeń obecnych modeli, ich rozwój i walidacja stanowią ważny krok w kierunku poprawy opieki nad pacjentami z COVID-19 poprzez umożliwienie wczesnej identyfikacji osób zagrożonych ciężkim przebiegiem choroby, co może przyczynić się do bardziej ukierunkowanych interwencji i potencjalnie lepszych wyników leczenia.

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.

Materiały źródłowe

#1
https://link.springer.com/article/10.1007/s10654-023-00973-x
Current evidence on COVID-19 prognostic models is inconsistent and clinical applicability remains controversial. […] We performed a systematic review to summarize and critically appraise the available studies that have developed, assessed and/or validated prognostic models of COVID-19 predicting health outcomes. […] Early identification of COVID-19 patients at risk of critical illness is crucial for early identification of patients requiring urgent medical attention or who would benefit the most from treatment. […] In addition, early prediction of the disease course not only enables cost-effective allocation of health care resources, but potentially decreases fatality rates as well. […] While several clinical prognostic models for COVID-19 have been described in the literature, they are limited in generalizability and/or applicability due to deficiencies in addressing fundamental statistical and methodological concerns.
#2 Outcome prediction model and prognostic biomarkers for COVID-19 patients in Vietnam
https://pmc.ncbi.nlm.nih.gov/articles/PMC9885243/
Accurate prognosis is important either after acute infection or during long-term follow-up of patients infected by severe acute respiratory syndrome coronavirus 2. This study aims to predict coronavirus disease 2019 (COVID-19) severity based on clinical and biological indicators, and to identify biomarkers for prognostic assessment. […] The random forest model could predict with 97% accuracy the likelihood of COVID-19 patients who subsequently worsened to the severe condition. The most important indicators were interleukin (IL)-6, ferritin and D-dimer. […] This study provided a simple and reliable model using two different sets of biomarkers to assess disease severity and predict clinical outcomes in COVID-19 patients in Vietnam. […] Our study results demonstrated that CRP, D-dimer, ferritin and IL-6 were significantly correlated with the prognosis of severe COVID-19 patients. Furthermore, this study provided a simple and reliable model to predict clinical outcomes for COVID-19 patients in Vietnam. Two different sets of biomarkers are applicable for the assessment of disease severity and prognosis.
#3 COVID-19 Outcome Prediction by Integrating Clinical and Metabolic Data using Machine Learning Algorithms
https://www.scielo.org.mx/scielo.php?script=sci_arttext&pid=S0034-83762022000600314
COVID-19 Outcome Prediction by Integrating Clinical and Metabolic Data using Machine Learning Algorithms […] Machine learning (ML) models could help physicians in identifying high-risk individuals. […] To study the use of ML models for COVID-19 prediction outcomes using clinical data and a combination of clinical and metabolic data, measured in a metabolomics facility from a public university. […] The model based on extended profile was more useful in early stages of the disease. Models based on clinical data were preferred for predicting severe and critical illness and death. […] ML and GAs provided adequate models to predict COVID-19 outcomes in patients with different severity grades. […] Since predicting the early outcomes of COVID-19 is challenging, machine learning (ML) models could help physicians in identifying high-risk individuals.
#4 Prediction models for diagnosis and prognosis of covid-19: systematic review and critical appraisal | The BMJ
https://www.bmj.com/content/369/bmj.m1328
Objective To review and appraise the validity and usefulness of published and preprint reports of prediction models for prognosis of patients with covid-19, and for detecting people in the general population at increased risk of covid-19 infection or being admitted to hospital or dying with the disease. […] Of these 731, 125 were diagnostic models (including 75 based on medical imaging) and the remaining 606 were prognostic models for either identifying those at risk of covid-19 in the general population (13 models) or predicting diverse outcomes in those individuals with confirmed covid-19 (593 models). […] The reported C indexes varied between 0.77 and 0.93 in development studies with low risk of bias, and between 0.56 and 0.78 in external validations with low risk of bias. […] Prediction models for covid-19 entered the academic literature to support medical decision making at unprecedented speed and in large numbers.
#5 Prediction models for diagnosis and prognosis of covid-19: systematic review and critical appraisal | The BMJ
https://www.bmj.com/content/369/bmj.m1328
Objective To review and appraise the validity and usefulness of published and preprint reports of prediction models for prognosis of patients with covid-19, and for detecting people in the general population at increased risk of covid-19 infection or being admitted to hospital or dying with the disease. […] Of these 731, 125 were diagnostic models (including 75 based on medical imaging) and the remaining 606 were prognostic models for either identifying those at risk of covid-19 in the general population (13 models) or predicting diverse outcomes in those individuals with confirmed covid-19 (593 models). […] The reported C indexes varied between 0.77 and 0.93 in development studies with low risk of bias, and between 0.56 and 0.78 in external validations with low risk of bias. […] Prediction models for covid-19 entered the academic literature to support medical decision making at unprecedented speed and in large numbers.
#6 Prediction models for diagnosis and prognosis of covid-19: systematic review and critical appraisal | The BMJ
https://www.bmj.com/content/369/bmj.m1328
Most published prediction model studies were poorly reported and at high risk of bias such that their reported predictive performances are probably optimistic. […] Models with low risk of bias should be validated before clinical implementation, preferably through collaborative efforts to also allow an investigation of the heterogeneity in their performance across various populations and settings. […] The 593 prognostic models for patients with covid-19 were more often at high risk of bias than the 13 general population models (90% (n=536) v 69% (n=9)). […] The median sample size for model development in patients with covid-19 was 397 (71 events), compared to 1.6 million (1867 events) for general population models. […] We identified 593 prognostic models for predicting clinical outcomes in patients with covid-19 (368 developments, 225 external validations).
#7 Prediction models for diagnosis and prognosis of covid-19: systematic review and critical appraisal | The BMJ
https://www.bmj.com/content/369/bmj.m1328
Objective To review and appraise the validity and usefulness of published and preprint reports of prediction models for prognosis of patients with covid-19, and for detecting people in the general population at increased risk of covid-19 infection or being admitted to hospital or dying with the disease. […] Of these 731, 125 were diagnostic models (including 75 based on medical imaging) and the remaining 606 were prognostic models for either identifying those at risk of covid-19 in the general population (13 models) or predicting diverse outcomes in those individuals with confirmed covid-19 (593 models). […] The reported C indexes varied between 0.77 and 0.93 in development studies with low risk of bias, and between 0.56 and 0.78 in external validations with low risk of bias. […] Prediction models for covid-19 entered the academic literature to support medical decision making at unprecedented speed and in large numbers.
#8
https://link.springer.com/article/10.1007/s10654-023-00973-x
Future large, multi-centric and well-designed prognostic prospective studies are needed to clarify remaining uncertainties. […] The most common outcome was mortality which was evaluated in 152 (48.4%) studies. […] Among studies that specified and reported prediction time for mortality there was variation in reporting. […] The difference between fixed time frame and time-to-event outcomes, lies mainly in the fact that time-to-event reporting takes account of whether an event takes place and the time at which the event occurs, such that both the event and the timing of the event are important. […] Predictors encountered most frequently in the 152 studies that developed or validated mortality prediction models were increased age, sex, decreased oxygen saturation, elevated levels of CRP, blood urea nitrogen, body temperature, number of comorbidities, unconsciousness, white blood cells count, lymphocyte count, D-dimer level, platelets and pulse rate.
#9
https://link.springer.com/article/10.1007/s10654-023-00973-x
Future large, multi-centric and well-designed prognostic prospective studies are needed to clarify remaining uncertainties. […] The most common outcome was mortality which was evaluated in 152 (48.4%) studies. […] Among studies that specified and reported prediction time for mortality there was variation in reporting. […] The difference between fixed time frame and time-to-event outcomes, lies mainly in the fact that time-to-event reporting takes account of whether an event takes place and the time at which the event occurs, such that both the event and the timing of the event are important. […] Predictors encountered most frequently in the 152 studies that developed or validated mortality prediction models were increased age, sex, decreased oxygen saturation, elevated levels of CRP, blood urea nitrogen, body temperature, number of comorbidities, unconsciousness, white blood cells count, lymphocyte count, D-dimer level, platelets and pulse rate.
#10
https://link.springer.com/article/10.1007/s10654-023-00973-x
Future large, multi-centric and well-designed prognostic prospective studies are needed to clarify remaining uncertainties. […] The most common outcome was mortality which was evaluated in 152 (48.4%) studies. […] Among studies that specified and reported prediction time for mortality there was variation in reporting. […] The difference between fixed time frame and time-to-event outcomes, lies mainly in the fact that time-to-event reporting takes account of whether an event takes place and the time at which the event occurs, such that both the event and the timing of the event are important. […] Predictors encountered most frequently in the 152 studies that developed or validated mortality prediction models were increased age, sex, decreased oxygen saturation, elevated levels of CRP, blood urea nitrogen, body temperature, number of comorbidities, unconsciousness, white blood cells count, lymphocyte count, D-dimer level, platelets and pulse rate.
#11
https://link.springer.com/article/10.1007/s10654-023-00973-x
Future large, multi-centric and well-designed prognostic prospective studies are needed to clarify remaining uncertainties. […] The most common outcome was mortality which was evaluated in 152 (48.4%) studies. […] Among studies that specified and reported prediction time for mortality there was variation in reporting. […] The difference between fixed time frame and time-to-event outcomes, lies mainly in the fact that time-to-event reporting takes account of whether an event takes place and the time at which the event occurs, such that both the event and the timing of the event are important. […] Predictors encountered most frequently in the 152 studies that developed or validated mortality prediction models were increased age, sex, decreased oxygen saturation, elevated levels of CRP, blood urea nitrogen, body temperature, number of comorbidities, unconsciousness, white blood cells count, lymphocyte count, D-dimer level, platelets and pulse rate.
#12
https://link.springer.com/article/10.1007/s10654-023-00973-x
Model performances assessed with AUC/ROC or c-index were reported in 133 studies and ranged between 0.49 to 0.99. […] The best reported predictive performance belonged to a model developed in Boston, USA, and externally validated in Wuhan, China, with 375 participants. […] Severity or critical illness in COVID-19 was reported in 66 (21.0%). […] The most frequently encountered predictors were older age, sex, body temperature, number of comorbidities (cardiovascular disease (CVD), hypertension, diabetes), decreased oxygen saturation, elevated levels of CRP, blood urea nitrogen (BUN), body temperature, systolic blood pressure (SBP), neutrophil-to-lymphocyte ratio (NLR), white blood cells count (WBC), lymphocyte count and pulse rate. […] A combined machine learning model (Support Vector Machine (SVM), Gradient Boosted Decision Tree (GBDT), and Neural Network(NN)) based on procalcitonin, [T+B+NK cell] count, IL-6, CRP, IL-2-receptor, T-helper lymphocyte/T-suppressor lymphocyte as predictors of critical illness had the best reported predictive performance, with an AUC of 0.99.
#13
https://link.springer.com/article/10.1007/s10654-023-00973-x
Model performances assessed with AUC/ROC or c-index were reported in 133 studies and ranged between 0.49 to 0.99. […] The best reported predictive performance belonged to a model developed in Boston, USA, and externally validated in Wuhan, China, with 375 participants. […] Severity or critical illness in COVID-19 was reported in 66 (21.0%). […] The most frequently encountered predictors were older age, sex, body temperature, number of comorbidities (cardiovascular disease (CVD), hypertension, diabetes), decreased oxygen saturation, elevated levels of CRP, blood urea nitrogen (BUN), body temperature, systolic blood pressure (SBP), neutrophil-to-lymphocyte ratio (NLR), white blood cells count (WBC), lymphocyte count and pulse rate. […] A combined machine learning model (Support Vector Machine (SVM), Gradient Boosted Decision Tree (GBDT), and Neural Network(NN)) based on procalcitonin, [T+B+NK cell] count, IL-6, CRP, IL-2-receptor, T-helper lymphocyte/T-suppressor lymphocyte as predictors of critical illness had the best reported predictive performance, with an AUC of 0.99.
#14
https://link.springer.com/article/10.1007/s10654-023-00973-x
Model performances assessed with AUC/ROC or c-index were reported in 133 studies and ranged between 0.49 to 0.99. […] The best reported predictive performance belonged to a model developed in Boston, USA, and externally validated in Wuhan, China, with 375 participants. […] Severity or critical illness in COVID-19 was reported in 66 (21.0%). […] The most frequently encountered predictors were older age, sex, body temperature, number of comorbidities (cardiovascular disease (CVD), hypertension, diabetes), decreased oxygen saturation, elevated levels of CRP, blood urea nitrogen (BUN), body temperature, systolic blood pressure (SBP), neutrophil-to-lymphocyte ratio (NLR), white blood cells count (WBC), lymphocyte count and pulse rate. […] A combined machine learning model (Support Vector Machine (SVM), Gradient Boosted Decision Tree (GBDT), and Neural Network(NN)) based on procalcitonin, [T+B+NK cell] count, IL-6, CRP, IL-2-receptor, T-helper lymphocyte/T-suppressor lymphocyte as predictors of critical illness had the best reported predictive performance, with an AUC of 0.99.
#15 Outcome prediction model and prognostic biomarkers for COVID-19 patients in Vietnam
https://pmc.ncbi.nlm.nih.gov/articles/PMC9885243/
Accurate prognosis is important either after acute infection or during long-term follow-up of patients infected by severe acute respiratory syndrome coronavirus 2. This study aims to predict coronavirus disease 2019 (COVID-19) severity based on clinical and biological indicators, and to identify biomarkers for prognostic assessment. […] The random forest model could predict with 97% accuracy the likelihood of COVID-19 patients who subsequently worsened to the severe condition. The most important indicators were interleukin (IL)-6, ferritin and D-dimer. […] This study provided a simple and reliable model using two different sets of biomarkers to assess disease severity and predict clinical outcomes in COVID-19 patients in Vietnam. […] Our study results demonstrated that CRP, D-dimer, ferritin and IL-6 were significantly correlated with the prognosis of severe COVID-19 patients. Furthermore, this study provided a simple and reliable model to predict clinical outcomes for COVID-19 patients in Vietnam. Two different sets of biomarkers are applicable for the assessment of disease severity and prognosis.
#16 Outcome prediction model and prognostic biomarkers for COVID-19 patients in Vietnam
https://pmc.ncbi.nlm.nih.gov/articles/PMC9885243/
Accurate prognosis is important either after acute infection or during long-term follow-up of patients infected by severe acute respiratory syndrome coronavirus 2. This study aims to predict coronavirus disease 2019 (COVID-19) severity based on clinical and biological indicators, and to identify biomarkers for prognostic assessment. […] The random forest model could predict with 97% accuracy the likelihood of COVID-19 patients who subsequently worsened to the severe condition. The most important indicators were interleukin (IL)-6, ferritin and D-dimer. […] This study provided a simple and reliable model using two different sets of biomarkers to assess disease severity and predict clinical outcomes in COVID-19 patients in Vietnam. […] Our study results demonstrated that CRP, D-dimer, ferritin and IL-6 were significantly correlated with the prognosis of severe COVID-19 patients. Furthermore, this study provided a simple and reliable model to predict clinical outcomes for COVID-19 patients in Vietnam. Two different sets of biomarkers are applicable for the assessment of disease severity and prognosis.
#17 Can we predict the severe course of COVID-19 – a systematic review and meta-analysis of indicators of clinical outcome? | PLOS One
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0255154
Can we predict the severe course of COVID-19 – a systematic review and meta-analysis of indicators of clinical outcome? […] In this systematic review and meta-analysis, we assessed demographic, laboratory and clinical indicators as predictors for severe courses of COVID-19. […] Age was strongly associated with mortality with a difference of medians (DoM) of 13.15 years (95% confidence interval (CI) 11.37 to 14.94) between those who died and those who survived. […] We found a clinically relevant difference between non-survivors and survivors for C-reactive protein (CRP; DoM 69.10 mg/L, CI 50.43 to 87.77), lactate dehydrogenase (LDH; DoM 189.49 U/L, CI 155.00 to 223.98), cardiac troponin I (cTnI; DoM 21.88 pg/mL, CI 9.78 to 33.99) and D-Dimer (DoM 1.29mg/L, CI 0.9 to 1.69). […] This comprehensive meta-analysis found age, cerebrovascular disease, CRP, LDH and cTnI to be the most important risk-factors that predict severe COVID-19 outcomes and will inform clinical scores to support early decision-making. […] Our data on mortality and ICU admission corroborate most of the proposed indicators of clinical outcomes, clarifies the strength of association and highlights additional indicators.
#18 Can we predict the severe course of COVID-19 – a systematic review and meta-analysis of indicators of clinical outcome? | PLOS One
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0255154
Can we predict the severe course of COVID-19 – a systematic review and meta-analysis of indicators of clinical outcome? […] In this systematic review and meta-analysis, we assessed demographic, laboratory and clinical indicators as predictors for severe courses of COVID-19. […] Age was strongly associated with mortality with a difference of medians (DoM) of 13.15 years (95% confidence interval (CI) 11.37 to 14.94) between those who died and those who survived. […] We found a clinically relevant difference between non-survivors and survivors for C-reactive protein (CRP; DoM 69.10 mg/L, CI 50.43 to 87.77), lactate dehydrogenase (LDH; DoM 189.49 U/L, CI 155.00 to 223.98), cardiac troponin I (cTnI; DoM 21.88 pg/mL, CI 9.78 to 33.99) and D-Dimer (DoM 1.29mg/L, CI 0.9 to 1.69). […] This comprehensive meta-analysis found age, cerebrovascular disease, CRP, LDH and cTnI to be the most important risk-factors that predict severe COVID-19 outcomes and will inform clinical scores to support early decision-making. […] Our data on mortality and ICU admission corroborate most of the proposed indicators of clinical outcomes, clarifies the strength of association and highlights additional indicators.
#19 Predicting the Outcome of Patients with Severe COVID-19 with Simple Inflammatory Biomarkers: The Utility of Novel Combined ScoresâResults from a European Tertiary/Referral Centre
https://www.mdpi.com/2077-0383/13/4/967
Results: […] Two points of C-CRP #1 and 2 points of C-CRP #3 predicted HFNC with a probability as high as 0.625 (p = 0.005) and 0.561 (p < 0.001), respectively. [...] For death, the optimal cut-off point for CRP was 1.11 and for CLR 3.2*1033. Two points of C-CRP #3* with an accuracy of 0.922 predicted mortality (p = 0.0038) in severe COVID-19. [...] Conclusions: The combination scores of CRP and inflammatory biomarkers, based on admission values, are promising predictors for respiratory support using HFNC and for mortality in patients suffering from severe COVID-19 infection.
#20 Predicting the Outcome of Patients with Severe COVID-19 with Simple Inflammatory Biomarkers: The Utility of Novel Combined ScoresâResults from a European Tertiary/Referral Centre
https://www.mdpi.com/2077-0383/13/4/967
Results: […] Two points of C-CRP #1 and 2 points of C-CRP #3 predicted HFNC with a probability as high as 0.625 (p = 0.005) and 0.561 (p < 0.001), respectively. [...] For death, the optimal cut-off point for CRP was 1.11 and for CLR 3.2*1033. Two points of C-CRP #3* with an accuracy of 0.922 predicted mortality (p = 0.0038) in severe COVID-19. [...] Conclusions: The combination scores of CRP and inflammatory biomarkers, based on admission values, are promising predictors for respiratory support using HFNC and for mortality in patients suffering from severe COVID-19 infection.
#21 Multimodal deep learning for COVID-19 prognosis prediction in the emergency department: a bi-centric study | Scientific Reports
https://www.nature.com/articles/s41598-023-37512-3
Predicting clinical deterioration in COVID-19 patients remains a challenging task in the Emergency Department (ED). […] The predicted outcomes were 30-day mortality and ICU admission. […] Our results suggest that a combined textual and tabular model can effectively predict COVID-19 prognosis which may assist ED physicians in their decision-making process. […] The main finding was that combining tabular and textual variables led to better prediction of 30-day mortality compared to models using only tabular variables. […] For ICU admission, the combined model had higher precision, specificity, F1 score, and MCC values. […] Therefore, integrating textual and tabular data may improve the prediction of negative outcomes in COVID-19 patients. […] The results of this study suggest that the combined analysis of tabular and textual data was effective in predicting 30-day mortality in patients with SARS-CoV-2 infection presenting to the ED and in predicting their ICU admission. […] Future studies will be necessary to improve the accuracy of the algorithm and confirm its generalizability through external validation cohorts, especially with the changing epidemiology of the pandemic.
#22 Multimodal deep learning for COVID-19 prognosis prediction in the emergency department: a bi-centric study | Scientific Reports
https://www.nature.com/articles/s41598-023-37512-3
Predicting clinical deterioration in COVID-19 patients remains a challenging task in the Emergency Department (ED). […] The predicted outcomes were 30-day mortality and ICU admission. […] Our results suggest that a combined textual and tabular model can effectively predict COVID-19 prognosis which may assist ED physicians in their decision-making process. […] The main finding was that combining tabular and textual variables led to better prediction of 30-day mortality compared to models using only tabular variables. […] For ICU admission, the combined model had higher precision, specificity, F1 score, and MCC values. […] Therefore, integrating textual and tabular data may improve the prediction of negative outcomes in COVID-19 patients. […] The results of this study suggest that the combined analysis of tabular and textual data was effective in predicting 30-day mortality in patients with SARS-CoV-2 infection presenting to the ED and in predicting their ICU admission. […] Future studies will be necessary to improve the accuracy of the algorithm and confirm its generalizability through external validation cohorts, especially with the changing epidemiology of the pandemic.
#23 Multimodal deep learning for COVID-19 prognosis prediction in the emergency department: a bi-centric study | Scientific Reports
https://www.nature.com/articles/s41598-023-37512-3
Predicting clinical deterioration in COVID-19 patients remains a challenging task in the Emergency Department (ED). […] The predicted outcomes were 30-day mortality and ICU admission. […] Our results suggest that a combined textual and tabular model can effectively predict COVID-19 prognosis which may assist ED physicians in their decision-making process. […] The main finding was that combining tabular and textual variables led to better prediction of 30-day mortality compared to models using only tabular variables. […] For ICU admission, the combined model had higher precision, specificity, F1 score, and MCC values. […] Therefore, integrating textual and tabular data may improve the prediction of negative outcomes in COVID-19 patients. […] The results of this study suggest that the combined analysis of tabular and textual data was effective in predicting 30-day mortality in patients with SARS-CoV-2 infection presenting to the ED and in predicting their ICU admission. […] Future studies will be necessary to improve the accuracy of the algorithm and confirm its generalizability through external validation cohorts, especially with the changing epidemiology of the pandemic.
#24 Multimodal deep learning for COVID-19 prognosis prediction in the emergency department: a bi-centric study | Scientific Reports
https://www.nature.com/articles/s41598-023-37512-3
Predicting clinical deterioration in COVID-19 patients remains a challenging task in the Emergency Department (ED). […] The predicted outcomes were 30-day mortality and ICU admission. […] Our results suggest that a combined textual and tabular model can effectively predict COVID-19 prognosis which may assist ED physicians in their decision-making process. […] The main finding was that combining tabular and textual variables led to better prediction of 30-day mortality compared to models using only tabular variables. […] For ICU admission, the combined model had higher precision, specificity, F1 score, and MCC values. […] Therefore, integrating textual and tabular data may improve the prediction of negative outcomes in COVID-19 patients. […] The results of this study suggest that the combined analysis of tabular and textual data was effective in predicting 30-day mortality in patients with SARS-CoV-2 infection presenting to the ED and in predicting their ICU admission. […] Future studies will be necessary to improve the accuracy of the algorithm and confirm its generalizability through external validation cohorts, especially with the changing epidemiology of the pandemic.
#25 Journal of Medical Internet Research – Prognostic Modeling of COVID-19 Using Artificial Intelligence in the United Kingdom: Model Development and Validation
https://www.jmir.org/2020/8/e20259/
Background: The current severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) outbreak is a public health emergency and the case fatality rate in the United Kingdom is significant. Although there appear to be several early predictors of outcome, there are no currently validated prognostic models or scoring systems applicable specifically to patients with confirmed SARS-CoV-2. […] Objective: We aim to create a point-of-admission mortality risk scoring system using an artificial neural network (ANN). […] Results: Patient-specific mortality was predicted with 86.25% accuracy, with a sensitivity of 87.50% (95% CI 61.65%-98.45%) and specificity of 85.94% (95% CI 74.98%-93.36%). The positive predictive value was 60.87% (95% CI 45.23%-74.56%), and the negative predictive value was 96.49% (95% CI 88.23%-99.02%). The area under the receiver operating characteristic curve was 90.12%.
#26 Journal of Medical Internet Research – Prognostic Modeling of COVID-19 Using Artificial Intelligence in the United Kingdom: Model Development and Validation
https://www.jmir.org/2020/8/e20259/
Conclusions: This analysis demonstrates an adaptive ANN trained on data at a single site, which demonstrates the early utility of deep learning approaches in a rapidly evolving pandemic with no established or validated prognostic scoring systems. […] Our aim was to provide a patient-specific, point-of-admission mortality risk prediction to help inform clinical management decisions at the earliest opportunity. The contribution of this analysis is in the proof-of-concept ANN trained on data from a single site, which demonstrates the early utility of deep learning approaches in a rapidly evolving pandemic with no established or validated prognostic scoring systems. […] Altered mentation, dyspnea, and increasing age were found to be the most salient overall features in predicting mortality. Moderate predictors of mortality included collapse, male gender, new cough, and previous respiratory pathology. These features are broadly in line with the current literature.
#27 Journal of Medical Internet Research – Prognostic Modeling of COVID-19 Using Artificial Intelligence in the United Kingdom: Model Development and Validation
https://www.jmir.org/2020/8/e20259/
An advantage of our current model is that all demographic, comorbidity, lifestyle, and symptom data can be collected on first encounter with a physician, and therefore an early outcome prediction can be produced following clerking. […] The ANN architecture is such that adding further input variables (including clinical investigation results) is easily achievable. We plan to extend the ANN in the future with these parameters to maximize its predictive capability. […] In the context of an evolving pandemic with no established prognostic scoring system, deep learning approaches can be used to rapidly develop empirical prognostic models. These models have the inherent advantage of becoming progressively more accurate and representative as data sets increase in size.
#28
https://scispace.com/papers/predicting-accident-severity-an-analysis-of-factors-3qwy3vpzks
COVID-19 Patient Health Prediction Using Boosted Random Forest Algorithm: A fine-tuned Random Forest model boosted by the AdaBoost algorithm is proposed that uses the COVID-19 patient’s geographical, travel, health, and demographic data to predict the severity of the case and the possible outcome, recovery, or death.
#29 Developing a COVID-19 mortality risk prediction model when individual-level data are not available | Nature Communications
https://www.nature.com/articles/s41467-020-18297-9
At the COVID-19 pandemic onset, when individual-level data of COVID-19 patients were not yet available, there was already a need for risk predictors to support prevention and treatment decisions. […] Here, we report a hybrid strategy to create such a predictor, combining the development of a baseline severe respiratory infection risk predictor and a post-processing method to calibrate the predictions to reported COVID-19 case-fatality rates. […] With the accumulation of a COVID-19 patient cohort, this predictor is validated to have good discrimination (area under the receiver-operating characteristics curve of 0.943) and calibration (markedly improved compared to that of the baseline predictor). […] We thus demonstrate that even at the onset of a pandemic, shrouded in epidemiologic fog of war, it is possible to provide a useful risk predictor, now widely used in a large healthcare organization.
#30 Developing a COVID-19 mortality risk prediction model when individual-level data are not available | Nature Communications
https://www.nature.com/articles/s41467-020-18297-9
At the COVID-19 pandemic onset, when individual-level data of COVID-19 patients were not yet available, there was already a need for risk predictors to support prevention and treatment decisions. […] Here, we report a hybrid strategy to create such a predictor, combining the development of a baseline severe respiratory infection risk predictor and a post-processing method to calibrate the predictions to reported COVID-19 case-fatality rates. […] With the accumulation of a COVID-19 patient cohort, this predictor is validated to have good discrimination (area under the receiver-operating characteristics curve of 0.943) and calibration (markedly improved compared to that of the baseline predictor). […] We thus demonstrate that even at the onset of a pandemic, shrouded in epidemiologic fog of war, it is possible to provide a useful risk predictor, now widely used in a large healthcare organization.
#31 Developing a COVID-19 mortality risk prediction model when individual-level data are not available | Nature Communications
https://www.nature.com/articles/s41467-020-18297-9
At the COVID-19 pandemic onset, when individual-level data of COVID-19 patients were not yet available, there was already a need for risk predictors to support prevention and treatment decisions. […] Here, we report a hybrid strategy to create such a predictor, combining the development of a baseline severe respiratory infection risk predictor and a post-processing method to calibrate the predictions to reported COVID-19 case-fatality rates. […] With the accumulation of a COVID-19 patient cohort, this predictor is validated to have good discrimination (area under the receiver-operating characteristics curve of 0.943) and calibration (markedly improved compared to that of the baseline predictor). […] We thus demonstrate that even at the onset of a pandemic, shrouded in epidemiologic fog of war, it is possible to provide a useful risk predictor, now widely used in a large healthcare organization.
#32 Developing a COVID-19 mortality risk prediction model when individual-level data are not available | Nature Communications
https://www.nature.com/articles/s41467-020-18297-9
The COVID-19 patient population used for validation included a cohort of 4179 COVID-19 patients that were diagnosed in CHS until April 16, 2020, with 143 (3.4%) deaths recorded until July 16, 2020. […] Discriminatory performance of the COVID-19 predictions in general and at specific thresholds is detailed in Table 2. […] The overall AUROC of the COVID-19 predictions is 0.943 (95% CI 0.9260.956). […] At a 5% risk threshold, a fraction of 15% (95% CI 1416%) of the testing population is found to be at high risk, with a sensitivity of 88% (95% CI 8393%) and PPV of 20% (95% CI 1723%). […] The sensitivity analysis for a composite outcome that considers both severe COVID-19 cases and COVID-19 mortality is presented in Supplementary Table 3. […] We showed that on a validation set of 4179 COVID-19 patients (3.4% death rate), the model was highly discriminative, with an AUROC of 0.943.
#33 The Predictive Performance of Risk Scores for the Outcome of COVID-19 in a 2-Year Swiss Cohort
https://www.mdpi.com/2227-9059/12/8/1702
Various scoring systems are available for COVID-19 risk stratification. This study aimed to validate their performance in predicting severe COVID-19 course in a large, heterogeneous Swiss cohort. Scores like the National Early Warning Score (NEWS), CURB-65, 4C mortality score (4C), Spanish Society of Infectious Diseases and Clinical Microbiology score (COVID-SEIMC), and COVID Intubation Risk Score (COVID-IRS) were assessed in patients hospitalized for COVID-19 in 2020 and 2021. Predictive accuracy for severe course (defined as all-cause in-hospital death or invasive mechanical ventilation (IMV)) was evaluated using receiver operating characteristic curves and the area under the curve (AUC). The new âCOVID-COMBIâ score, combining parameters from the top two scores, was also validated. This study included 1,051 patients (mean age 65 years, 60% male), with 162 (15%) experiencing severe course. Among the established scores, 4C had the best accuracy for predicting severe course (AUC 0.76), followed by COVID-IRS (AUC 0.72). COVID-COMBI showed significantly higher accuracy than all established scores (AUC 0.79, p = 0.001). For predicting in-hospital death, 4C performed best (AUC 0.83), and, for IMV, COVID-IRS performed best (AUC 0.78). The 4C and COVID-IRS scores were robust predictors of severe COVID-19 course, while the new COVID-COMBI showed significantly improved accuracy but requires further validation.
#34 The Predictive Performance of Risk Scores for the Outcome of COVID-19 in a 2-Year Swiss Cohort
https://www.mdpi.com/2227-9059/12/8/1702
Various scoring systems are available for COVID-19 risk stratification. This study aimed to validate their performance in predicting severe COVID-19 course in a large, heterogeneous Swiss cohort. Scores like the National Early Warning Score (NEWS), CURB-65, 4C mortality score (4C), Spanish Society of Infectious Diseases and Clinical Microbiology score (COVID-SEIMC), and COVID Intubation Risk Score (COVID-IRS) were assessed in patients hospitalized for COVID-19 in 2020 and 2021. Predictive accuracy for severe course (defined as all-cause in-hospital death or invasive mechanical ventilation (IMV)) was evaluated using receiver operating characteristic curves and the area under the curve (AUC). The new âCOVID-COMBIâ score, combining parameters from the top two scores, was also validated. This study included 1,051 patients (mean age 65 years, 60% male), with 162 (15%) experiencing severe course. Among the established scores, 4C had the best accuracy for predicting severe course (AUC 0.76), followed by COVID-IRS (AUC 0.72). COVID-COMBI showed significantly higher accuracy than all established scores (AUC 0.79, p = 0.001). For predicting in-hospital death, 4C performed best (AUC 0.83), and, for IMV, COVID-IRS performed best (AUC 0.78). The 4C and COVID-IRS scores were robust predictors of severe COVID-19 course, while the new COVID-COMBI showed significantly improved accuracy but requires further validation.
#35 The Predictive Performance of Risk Scores for the Outcome of COVID-19 in a 2-Year Swiss Cohort
https://www.mdpi.com/2227-9059/12/8/1702
Even though the SARS-CoV-2 pandemic has strongly subsided, the identification of risk factors and the prediction of a severe course remain important. The early identification of individuals at high risk can prevent the progression of COVID-19 to ARDS by allowing physicians to initiate prompt interventions and appropriate treatment strategies. Additionally, the prediction of a severe course can provide valuable prognostic information for healthcare providers, affected patients, and their relatives. Finally, the identification of risk factors for a severe course of COVID-19 helps define suitable eligibility criteria for clinical trials further investigating preventive and therapeutic options for COVID-19. […] Our study demonstrates that both the 4C mortality score and the COVID-IRS score have robust predictive accuracy for the prediction of severe course in patients hospitalized for COVID-19. Notably, the newly developed COVID-COMBI score outperformed both scores in the prediction of severe disease progression. Additionally, the COVID-COMBI score performed well in the prediction of in-hospital death and invasive mechanical ventilation. The findings from this study highlight the potential utility of the COVID-COMBI score in clinical practice. By integrating comprehensive risk factors and clinical routine parameters, the COVID-COMBI can help healthcare providers identify high-risk patients early, enabling timely interventions and potentially improving patient outcomes. The scoreâs robust performance across various outcomes indicates its practical value in managing COVID-19 patients. Future research should focus on the prospective and multicentric external validation of COVID-COMBI to confirm its predictive accuracy in diverse healthcare settings.
#36 Prediction models for diagnosis and prognosis of covid-19: systematic review and critical appraisal | The BMJ
https://www.bmj.com/content/369/bmj.m1328
Most published prediction model studies were poorly reported and at high risk of bias such that their reported predictive performances are probably optimistic. […] Models with low risk of bias should be validated before clinical implementation, preferably through collaborative efforts to also allow an investigation of the heterogeneity in their performance across various populations and settings. […] The 593 prognostic models for patients with covid-19 were more often at high risk of bias than the 13 general population models (90% (n=536) v 69% (n=9)). […] The median sample size for model development in patients with covid-19 was 397 (71 events), compared to 1.6 million (1867 events) for general population models. […] We identified 593 prognostic models for predicting clinical outcomes in patients with covid-19 (368 developments, 225 external validations).
#37 Prediction models for diagnosis and prognosis of covid-19: systematic review and critical appraisal | The BMJ
https://www.bmj.com/content/369/bmj.m1328
Most published prediction model studies were poorly reported and at high risk of bias such that their reported predictive performances are probably optimistic. […] Models with low risk of bias should be validated before clinical implementation, preferably through collaborative efforts to also allow an investigation of the heterogeneity in their performance across various populations and settings. […] The 593 prognostic models for patients with covid-19 were more often at high risk of bias than the 13 general population models (90% (n=536) v 69% (n=9)). […] The median sample size for model development in patients with covid-19 was 397 (71 events), compared to 1.6 million (1867 events) for general population models. […] We identified 593 prognostic models for predicting clinical outcomes in patients with covid-19 (368 developments, 225 external validations).
#38
https://link.springer.com/article/10.1007/s10654-023-00973-x
Overall, the RoB assessment was rated to be at high or unclear RoB in 312 studies. […] This could be explained by shortcomings such as poor methodological quality, small sample size, poor handling of missing data, failure to deal with overfitting, definitions of COVID-19 based on clinical features rather than on the result of laboratory diagnostic test for SARS-CoV-2 and its severity with studies using heterogeneous outcomes. […] Future large, multi-center and well-designed prospective studies are needed for the development of predictive models for COVID-19 with clinical utility that can be applied to diverse populations.
#39
https://link.springer.com/article/10.1007/s10654-023-00973-x
Overall, the RoB assessment was rated to be at high or unclear RoB in 312 studies. […] This could be explained by shortcomings such as poor methodological quality, small sample size, poor handling of missing data, failure to deal with overfitting, definitions of COVID-19 based on clinical features rather than on the result of laboratory diagnostic test for SARS-CoV-2 and its severity with studies using heterogeneous outcomes. […] Future large, multi-center and well-designed prospective studies are needed for the development of predictive models for COVID-19 with clinical utility that can be applied to diverse populations.
#40 COVID 19 â No one can precisely predict the outcome, that includes Data Models
https://www.linkedin.com/pulse/covid-19-one-can-precisely-predict-outcome-includes-data-singh
No one can precisely predict the outcome, that includes Data Models. […] The outputs from any/all these models can vary by a BIG amount by a slight change in assumption(s) which is basically are input assumptions given to the model for a desired output. […] But in the end, there is still no telling one might see another high noon come standoff.
#41 COVID 19 â No one can precisely predict the outcome, that includes Data Models
https://www.linkedin.com/pulse/covid-19-one-can-precisely-predict-outcome-includes-data-singh
No one can precisely predict the outcome, that includes Data Models. […] The outputs from any/all these models can vary by a BIG amount by a slight change in assumption(s) which is basically are input assumptions given to the model for a desired output. […] But in the end, there is still no telling one might see another high noon come standoff.
#42
https://link.springer.com/article/10.1007/s10654-023-00973-x
Current evidence on COVID-19 prognostic models is inconsistent and clinical applicability remains controversial. […] We performed a systematic review to summarize and critically appraise the available studies that have developed, assessed and/or validated prognostic models of COVID-19 predicting health outcomes. […] Early identification of COVID-19 patients at risk of critical illness is crucial for early identification of patients requiring urgent medical attention or who would benefit the most from treatment. […] In addition, early prediction of the disease course not only enables cost-effective allocation of health care resources, but potentially decreases fatality rates as well. […] While several clinical prognostic models for COVID-19 have been described in the literature, they are limited in generalizability and/or applicability due to deficiencies in addressing fundamental statistical and methodological concerns.
#43
https://link.springer.com/article/10.1007/s10654-023-00973-x
Overall, the RoB assessment was rated to be at high or unclear RoB in 312 studies. […] This could be explained by shortcomings such as poor methodological quality, small sample size, poor handling of missing data, failure to deal with overfitting, definitions of COVID-19 based on clinical features rather than on the result of laboratory diagnostic test for SARS-CoV-2 and its severity with studies using heterogeneous outcomes. […] Future large, multi-center and well-designed prospective studies are needed for the development of predictive models for COVID-19 with clinical utility that can be applied to diverse populations.
#44 Comparing visual outcomes of nAMD treatment during and after the COVID-19 restrictions period | PLOS One
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0323253
Compare treatment outcomes of newly diagnosed neovascular age-related macular degeneration (nAMD) during and after the COVID-19 restrictions. […] Patients diagnosed during the COVID-19 restrictions period had worse visual outcomes than those diagnosed thereafter. Multiple factors, including, but not limited to reduced treatment frequency, likely contributed to worse visual outcomes. […] The pandemic was associated with poorer visual outcomes for patients diagnosed during the restricted period (Group 1) compared to those diagnosed after this period (Group 2). […] Patients in Group 1 had worse visual outcomes at 12 months and last follow-up compared to Group 2, with a lower proportion achieving a BCVA20/40 and a higher proportion with BCVA20/70. […] Our findings emphasize the importance of developing adaptable strategies to maintain effective patient care in the face of external challenges posed by a global pandemic.
#45 Comparing visual outcomes of nAMD treatment during and after the COVID-19 restrictions period | PLOS One
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0323253
Compare treatment outcomes of newly diagnosed neovascular age-related macular degeneration (nAMD) during and after the COVID-19 restrictions. […] Patients diagnosed during the COVID-19 restrictions period had worse visual outcomes than those diagnosed thereafter. Multiple factors, including, but not limited to reduced treatment frequency, likely contributed to worse visual outcomes. […] The pandemic was associated with poorer visual outcomes for patients diagnosed during the restricted period (Group 1) compared to those diagnosed after this period (Group 2). […] Patients in Group 1 had worse visual outcomes at 12 months and last follow-up compared to Group 2, with a lower proportion achieving a BCVA20/40 and a higher proportion with BCVA20/70. […] Our findings emphasize the importance of developing adaptable strategies to maintain effective patient care in the face of external challenges posed by a global pandemic.
#46 Comparing visual outcomes of nAMD treatment during and after the COVID-19 restrictions period | PLOS One
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0323253
Compare treatment outcomes of newly diagnosed neovascular age-related macular degeneration (nAMD) during and after the COVID-19 restrictions. […] Patients diagnosed during the COVID-19 restrictions period had worse visual outcomes than those diagnosed thereafter. Multiple factors, including, but not limited to reduced treatment frequency, likely contributed to worse visual outcomes. […] The pandemic was associated with poorer visual outcomes for patients diagnosed during the restricted period (Group 1) compared to those diagnosed after this period (Group 2). […] Patients in Group 1 had worse visual outcomes at 12 months and last follow-up compared to Group 2, with a lower proportion achieving a BCVA20/40 and a higher proportion with BCVA20/70. […] Our findings emphasize the importance of developing adaptable strategies to maintain effective patient care in the face of external challenges posed by a global pandemic.
#47 Multimodal deep learning for COVID-19 prognosis prediction in the emergency department: a bi-centric study | Scientific Reports
https://www.nature.com/articles/s41598-023-37512-3
Predicting clinical deterioration in COVID-19 patients remains a challenging task in the Emergency Department (ED). […] The predicted outcomes were 30-day mortality and ICU admission. […] Our results suggest that a combined textual and tabular model can effectively predict COVID-19 prognosis which may assist ED physicians in their decision-making process. […] The main finding was that combining tabular and textual variables led to better prediction of 30-day mortality compared to models using only tabular variables. […] For ICU admission, the combined model had higher precision, specificity, F1 score, and MCC values. […] Therefore, integrating textual and tabular data may improve the prediction of negative outcomes in COVID-19 patients. […] The results of this study suggest that the combined analysis of tabular and textual data was effective in predicting 30-day mortality in patients with SARS-CoV-2 infection presenting to the ED and in predicting their ICU admission. […] Future studies will be necessary to improve the accuracy of the algorithm and confirm its generalizability through external validation cohorts, especially with the changing epidemiology of the pandemic.
#48 Journal of Medical Internet Research – Prognostic Modeling of COVID-19 Using Artificial Intelligence in the United Kingdom: Model Development and Validation
https://www.jmir.org/2020/8/e20259/
An advantage of our current model is that all demographic, comorbidity, lifestyle, and symptom data can be collected on first encounter with a physician, and therefore an early outcome prediction can be produced following clerking. […] The ANN architecture is such that adding further input variables (including clinical investigation results) is easily achievable. We plan to extend the ANN in the future with these parameters to maximize its predictive capability. […] In the context of an evolving pandemic with no established prognostic scoring system, deep learning approaches can be used to rapidly develop empirical prognostic models. These models have the inherent advantage of becoming progressively more accurate and representative as data sets increase in size.