Informative Ranking of Stand Out Collections of Symptoms: A New Data-Driven Approach to Identify the Strong Warning Signs of COVID 19
AlMomani, A. A. R.; Bollt, E. M.
Show abstract
We develop here a data-driven approach for disease recognition based on given symptoms, to be efficient tool for anomaly detection. In a clinical setting and when presented with a patient with a combination of traits, a doctor may wonder if a certain combination of symptoms may be especially predictive, such as the question, "Are fevers more informative in women than men?" The answer to this question is, yes. We develop here a methodology to enumerate such questions, to learn what are the stronger warning signs when attempting to diagnose a disease, called Conditional Predictive Informativity, (CPI), whose ranking we call CPIR. This simple to use process allows us to identify particularly informative combinations of symptoms and traits that may help medical field analysis in general, and possibly to become a new data-driven advised approach for individual medical diagnosis, as well as for broader public policy discussion. In particular we have been motivated to develop this tool in the current environment of the pressing world crisis due to the COVID 19 pandemic. We apply the methods here to data collected from national, provincial, and municipal health reports, as well as additional information from online, and then curated to an online publically available Github repository.
Matching journals
The top 12 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Measuring COVID-19 and Influenza in the Real World via Person-Generated Health Data 93%
- Building a Best-in-Class De-identification Tool for Electronic Medical Records Through Ensemble Learning 92%
- Inferring global-scale temporal latent topics from news reports to predict public health interventions for COVID-19 92%
Similar papers in this journal
Similar papers in this journal
- Achieving Inclusive Healthcare through Integrating Education and Research with AI and Personalized Curricula 91%
- Forecasting hospital-level COVID-19 admissions using real-time mobility data 91%
- Extraction of Crohn's Disease Clinical Phenotypes from Clinical Text Using Natural Language Processing 91%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.