General practitioners are the first line of defense against COVID-19. TADA helps them diagnose infected patients with satisfying accuracy.

Industry

Healthcare – Medical Research

Project Duration and Effort

Two weeks

Customer Benefits

  • 88% accuracy in supporting medical diagnosis with less than a 100 records for training data
  • Results obtained in 2 weeks instead of 6 months
  • Problem to solve

    A group of fifteen general practitioners based in France have gathered less than a hundred anonymized medical records since the outbreak of the COVID-19 infection. They want to find an efficient and reliable way to diagnose COVID-19 infected patients. COVID-19 being a multifaceted disease, it is hard to identify which symptoms are crucial. Their goal is to gain time in making their diagnoses. Which, in turn, enables general practitioners to improve patients’ care.

    Overview of TADA interface

    Objectives

    • With less than a hundred records for training, accurately identify patients at risk of being infected.
    • Spot at most three significant redline symptoms.
    Overview of the global influence
    It raises the following general medicine question: Can we predict with precision who is sick with COVID-19 with two or three symptoms?

    Solution

    After training TADA with less than 100 patients records, containing data on symptoms, age, and gender, we obtained a prediction of infected people with an 88% accuracy.

    TADA helped to identify the essential three symptoms, i.e., stiffness, chest pain and shortness of breath. It took two weeks to develop this analysis instead of six months. The accuracy is excellent, with 88%.

    Customer Benefits

    In two weeks, general practitioners gained a competitive edge by:
    • Distinguishing essential symptoms to recognize COVID-19 infection: stiffness, chest pain and shortness of breath,
    • Produce models in 2 weeks vs. six months while maintaining accuracy at 88%.