Platform

Every risk score traces back to real signals.

Keel isn't a black-box score. It's a transparent pipeline from raw signal to reviewable, explainable recommendation — built for teams who have to defend every decision they make.

Signal fusion

Five data domains, one patient timeline.

Most adherence tools only look at claims — which is exactly why they catch problems weeks too late. Keel fuses the signals that actually explain why a patient drifts.

Pharmacy claims

Fill history, days-supply, switch and discontinuation patterns across pharmacy and medical benefit.

Specialty pharmacy activity

Dispense holds, shipment status, refill outreach attempts and responses.

Benefit events

Formulary changes, prior authorizations, cost-share resets, coverage transitions.

Clinical records

Visit notes, labs, disease activity scores, documented tolerability issues.

Care-management interactions

Call notes, intake updates, life-stage and social-barrier flags from the care team.

Explainability

No opaque scores. Ever.

Every risk score decomposes into the individual signals that produced it, each with its own weight and evidence. Pharmacists see exactly what changed and when — so a score is never just a number to trust blindly.

Example patient
Diane OkaforRisk 87

Adalimumab (Humira biosimilar) · Rheumatology

Contributing signals
BenefitFormulary substitution
+34

Payer moved preferred tier from originator to biosimilar effective May 18 — new prior authorization required.

ClaimsRefill gap beyond grace window
+29

No claim submitted 11 days after the 5-day grace period closed.

Specialty pharmacyOutreach attempts unanswered
+15

Specialty pharmacy logged 2 automated refill reminders with no response.

ClinicalRising disease activity signal
+9

Most recent CDAI documented at moderate activity, up from low at last visit.

Recommendation

A specific next step, matched to the reason for drift.

Keel doesn’t stop at “at risk.” It classifies the likely driver — cost, access, logistics, tolerability, or clinical concern — and recommends the channel most likely to resolve it.

Phone outreach

For concerns that need a clinical voice — safety questions, tolerability, motivation.

Secure message

For low-friction nudges to patients who respond well to asynchronous contact.

Provider referral

For clinical holds and monitoring gaps that need direct coordination with the prescriber.

Benefits navigation

For cost- and coverage-driven gaps — copay programs, PA status, plan transitions.

Marcus Webb
Palbociclib · Oncology · PT-10517
High risk
Recommended intervention
Benefits navigation + manufacturer copay program screen

Copay rose 13.6x after benefit reset. Pattern matches cost-driven abandonment — patient likely qualifies for manufacturer assistance.

Model confidence: 88%
Approve
Modify
Dismiss
Reviewed by Priya Nair, PharmD — decision and rationale logged to patient record.
Pharmacist review

One queue. One decision at a time.

Every flagged patient lands in a single review queue with the full evidence trail attached. A pharmacist approves the recommendation, modifies the approach, or dismisses the flag with a note — every decision is logged back to the patient record and feeds future recommendations.

FAQ

Common questions from clinical teams.

Does Keel make clinical decisions on its own?

No. Keel surfaces risk, evidence, and a recommended next step. A licensed pharmacist reviews and decides on every case — approve, modify, or dismiss. Keel never contacts a patient or changes a care plan autonomously.

How is the risk score calculated?

Keel combines weighted signals from five data domains — claims, specialty pharmacy, benefit, clinical, and care management — into a single 0–100 score. Every score is fully traceable to the underlying signals and their individual weights, shown directly in the review interface.

How fresh is the underlying data?

Specialty pharmacy and benefit signals typically refresh within hours. Medical and pharmacy claims follow standard payer feed cadences, usually 24–72 hours, which is still materially faster than traditional claims-lag reporting.

Can we tune what counts as high risk?

Yes. Thresholds, signal weights, and routing rules are configurable per therapeutic area and per program, in partnership with your clinical and analytics teams during onboarding.

See the queue your team would be working today.

Walk through a live risk queue and a full patient review, with the same explainable risk scoring your team would work from.