Machine learning approach to dynamic risk modeling of mortality in COVID-19: a UK Biobank study

Dabbah, Mohammad A.; Reed, Angus B.; Booth, Adam T.C.; Yassaee, Arrash; Despotovic, Alex; Klasmer, Benjamin; Binning, Emily; Aral, Mert; Plans, David; Labrique, Alain; Mohan, Diwakar

doi:10.1101/2021.02.08.21251343

Cited by 3 publications

(1 citation statement)

References 63 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Using machine learning models leveraging this vital sign data, early detection models can be designed, including for SARS-CoV-2. While a number of risk prediction or prognostic prediction models have been developed for SARS-CoV-2 14,15,22,24 , there are a much smaller number of published diagnostic models.…”

Section: Introductionmentioning

confidence: 99%

Prediction of SARS-CoV-2 PCR positivity in UK Fenland study participants using smartphone-submitted vital sign measurements

Moore¹,

Doležalová²,

Morelli³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

IntroductionMobile health applications are increasingly being used in health and clinical research. SARS-CoV-2 has proven to have high infectivity, making outbreaks difficult to contain. Early detection can help prevent spread, but there is a need to develop easy-to-use screening tools that can help identify potential infection as early as possible. Here, we describe the development of a machine learning classifier that can predict SARS-CoV-2 PCR positivity using smartphone-submitted vital sign measurements.MethodsThe Fenland App study followed 2,199 UK participants using a smartphone application from August 2020 and for a minimum of six months. Participants completed a baseline questionnaire and then monthly questionnaires about SARS-CoV-2 status and vaccinations. Three times a week, participants provided measurements of their blood oxygen saturation, body temperature, and resting heart rate via a pulse oximeter, digital thermometer, and their smartphone. The participants participated in self initiated SARS-CoV-2 testing as per concurrent public health guidelines.We built predictive models SARS-CoV-2 PCR positivity status as obtained from national surveillance PCR test results.ResultsA total of 77 positive and 6,339 negative SARS-CoV-2 tests were recorded during the study. The final model achieved an ROC AUC of 0.695 ± 0.045. There was no difference in model performance when using 4, 8 or 12 weeks of baseline data before a SARS-CoV-2 test (F(2) = 0.80, p = 0.472). Addition of demographic or symptom information had no impact on model performance.ConclusionsUsing only three smartphone collected vital sign measurements, it is possible to predict SARS-CoV-2 PCR positivity, using a four week baseline period. Smartphone based remote monitoring of patient vital signs could allow for earlier screening for potential infections. This method could be applicable to any infectious disease that causes physiological changes in vital signs.

show abstract

Section: Introductionmentioning

confidence: 99%

Prediction of SARS-CoV-2 PCR positivity in UK Fenland study participants using smartphone-submitted vital sign measurements

Moore¹,

Doležalová²,

Morelli³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

Review of Machine Learning-Based Disease Diagnosis and Severity Estimation of COVID-19

Ayyavaraiah¹,

Venkateswarlu²

2023

Advances in Intelligent Systems and Computing

View full text Add to dashboard Cite

On Comparing Cross-Validated Forecasting Models with a Novel Fuzzy-TOPSIS Metric: A COVID-19 Case Study

et al. 2021

View full text Add to dashboard Cite

Time series cross-validation is a technique to select forecasting models. Despite the sophistication of cross-validation over single test/training splits, traditional and independent metrics, such as Mean Absolute Error (MAE) and Root Mean Square Error (RMSE), are commonly used to assess the model’s accuracy. However, what if decision-makers have different models fitting expectations to each moment of a time series? What if the precision of the forecasted values is also important? This is the case of predicting COVID-19 in Amapá, a Brazilian state in the Amazon rainforest. Due to the lack of hospital capacities, a model that promptly and precisely responds to notable ups and downs in the number of cases may be more desired than average models that only have good performances in more frequent and calm circumstances. In line with this, this paper proposes a hybridization of the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) and fuzzy sets to create a similarity metric, the closeness coefficient (CC), that enables relative comparisons of forecasting models under heterogeneous fitting expectations and also considers volatility in the predictions. We present a case study using three parametric and three machine learning models commonly used to forecast COVID-19 numbers. The results indicate that the introduced fuzzy similarity metric is a more informative performance assessment metric, especially when using time series cross-validation.

show abstract

Machine learning approach to dynamic risk modeling of mortality in COVID-19: a UK Biobank study

Cited by 3 publications

References 63 publications

Prediction of SARS-CoV-2 PCR positivity in UK Fenland study participants using smartphone-submitted vital sign measurements

Prediction of SARS-CoV-2 PCR positivity in UK Fenland study participants using smartphone-submitted vital sign measurements

Review of Machine Learning-Based Disease Diagnosis and Severity Estimation of COVID-19

On Comparing Cross-Validated Forecasting Models with a Novel Fuzzy-TOPSIS Metric: A COVID-19 Case Study

Contact Info

Product

Resources

About