Predict risk. Prevent it.
Clinical-grade prediction models that surface the right member, at the right moment, in the right workflow - peer-reviewed, transparent, and built for action. Not another opaque risk score.
Six clinical-grade models. Every member, every day.
Each model is peer-reviewed, continuously revalidated, and wired into the workflows where the prediction actually matters.
ED Admission Risk (90-day)
Predicts likelihood of an emergency department visit in the next 90 days, with primary driver explanations.
Inpatient Hospitalization
30/60/90-day inpatient admission likelihood, with avoidable-admit segmentation and intervention recommendations.
Total Cost Trajectory
Forward-looking 12-month cost forecast with rising-cost flags 3–6 months before traditional models notice.
Rising Risk / Disease Onset
Identifies pre-clinical signals of CKD, CHF, COPD, T2D progression - moving members from reactive to proactive care.
Readmission Risk (30-day)
Discharge-event–triggered readmission probability with care-transition recommendations.
Adherence & Engagement
Predicts medication non-adherence, missed appointments, and outreach responsiveness - feeds Maxine's call cadences.
A different kind of risk score.
Most risk models were built for population-level reporting, not for triggering action on a specific member tomorrow morning. Ours were built for action from day one.
- Clinical-grade. Co-developed with physicians; peer-reviewed in NEJM Catalyst, JMIR, and the Lancet Digital Health.
- Transparent. Every prediction comes with the top drivers and a clinical narrative - no black boxes.
- Real-time. Refreshes as new claims, EHR, Rx, and lab data arrive - not on a quarterly batch.
- Actionable. Pipes directly into agent workflows, Maxine call lists, care management worklists, and your dashboards.
- Audit-ready. Full lineage, fairness monitoring, drift detection. NIST AI RMF aligned.
Predictions only matter when they show up in the workflow.
Inside Maxine
Predictions prioritize Maxine's call cadences - the highest-impact members reached first, with the right script.
Care management worklists
Risk-tiered lists in your CM platform with explanations clinicians can act on. Not just a score in a CSV.
Agent decisioning
Prior auth and triage agents use predictions to weight clinical risk and route appropriately.
Your dashboards & data warehouse
Streamed to Snowflake / Databricks via API. Plug into your BI tools. Build your own apps on top.
See predictions running on your population.
Send us a de-identified claims sample. We'll run our model library and walk you through actionable cohorts in two weeks.
Start a model evaluation