WhyRiskRisk, explainedCreate your account

Explainable healthcare ML

Know your risk.
See exactly why.

WhyRisk estimates your diabetes risk with a trained machine-learning model — and shows you the reasons behind the score, in plain language. No black box. No invented numbers.

Create your account How the model works
Honest by design: one trained model today (diabetes). Everything else is clearly marked as roadmap — this site never dresses up a guess as a prediction.
Patient · Diabetes riskDemo data

Ahmed Raza, 47

Sample assessment · LightGBM model

0.31 / 1.00 probability
LOWMODERATEHIGH

Why this score — top factors

BMI 29.4+0.14
High blood pressure+0.10
High cholesterol+0.08
Physically active−0.06

Explanation generated with SHAP from the patient's own inputs. A screening aid — not a diagnosis.

0.8307AUC-ROC of the diabetes model on a held-out test split — the only accuracy figure we claim
1Trained model in production. Six more conditions are on the roadmap, labelled as such
21Health indicators per assessment, from the CDC BRFSS 2015 dataset (253,680 records)
SHAPPredictions ship with a SHAP explanation — each score shows the factors behind it

What it does

Three things, done properly — instead of ten things done with smoke and mirrors.

Risk prediction

A LightGBM model trained on CDC survey data estimates diabetes risk from 21 everyday health indicators — no lab visit needed for a first screening.

The reason, not just the score

Every result lists the factors that pushed the score up or down, with their weights. A doctor can argue with it — that is the point.

Records & appointments

Patient and doctor portals with role-based access, appointment booking and an admin overview — protected by row-level security.

What is real, what is next

Most health apps hide this page. We put it on the homepage.

Available now

  • Diabetes risk model (LightGBM)Trained
  • SHAP explanationsStored per prediction
  • Symptom screeningRule-based · labelled
  • Patient / doctor portals & appointmentsLive

On the roadmap

  • Heart disease modelIn development
  • Stroke modelIn development
  • Chronic kidney disease modelIn development
  • Hypertension modelIn development
  • Asthma & COPD modelsIn development

Inside the app

One admin, one login — a live overview of the whole practice.

Admin · Overview
—Total patients (live count)
—Total doctors (live count)
—Pending approvals
—Appointments

Counts load live from the database — the preview leaves them blank rather than inventing numbers.