Identifying Biomarkers in Diabetic Patients Using a Machine Learning Model for a Digital Therapeutics Organization

Explore how Reveal used a machine learning (ML) model to identify biomarkers for a digital therapeutics organization.

Identifying the Biomarkers

Our client wanted Reveal’s support to identify biomarkers of patients who are likely to respond to a treatment. Knowing these biomarkers would significantly improve patient outcomes. However, the challenge was that every patient is genetically unique and identifying biomarkers responsible for a good prognosis is not easy.

Revolution in Digital Therapeutics

Reveal developed an ML model that could predict the A1C (average blood sugar levels) changes at stipulated periods using the client’s digital therapeutic. The model integrated data from several sources such as health trackers, clinical systems, apps used by the patient, self-reported surveys, etc. It was layered with a provider-facing app to organize the provider’s patient panel for optimal triage usage. Additionally, the solution used observational ML models to inform HEOR and identify target patient populations who are likely to respond to our client’s digital therapeutic.

Impact Achieved

The insights gathered from Reveal’s solution supported a strategic partnership between the payer and the client. They jointly published a paper on biomarker development using machine learning, titled “Emergence of digital biomarkers to predict and modify treatment efficacy: machine learning study.”