SHAARPEC Foresight models patient trajectories at population scale — surfacing where demand is forming before it appears in conventional data.
Intelligence for commercial teams. Evidence for HEOR and Medical Affairs. One platform.
The gap shows up differently depending on where you sit.
54% of pharma launches underperformed against forecast between 2020 and 2023. The signals that predict access friction — prior auth, affordability, site-of-care setup — appear in claims data before launch. Most analytics don't read them.
Trinity Life Sciences, Moving the Needle: Lessons from the 2023 Launch Class, 2024
Real-world evidence studies age out within months of publication. The data to run them continuously already exists in most organizations — the infrastructure to do it doesn't.
Purpose-built for how patients move through care over time.
Every encounter, diagnosis, referral, and intervention is a connected event in time — not a row in a table. This structure is what makes it possible to detect patterns that flat claims analytics cannot see.
One disease-specific model learns the full range of how patients with a given condition move through care — nationally, across hundreds of thousands of trajectories. Every predictive algorithm built on that foundation inherits its depth, rather than being trained from scratch.
Conventional analytics describe what already happened. SHAARPEC Foresight was architected from the ground up to predict what's forming next — generating signal from clinical events that precede treatment, not from downstream prescription data.
US engagements begin with your existing claims data — the foundation of every trajectory model. The platform is architected to integrate EHR, SDOH, lab, and registry data as those sources become available, deepening prediction accuracy over time without rebuilding from scratch.
This is not propensity scoring dressed up as prediction. It is not statistical extrapolation of last quarter's prescription data. SHAARPEC Foresight trains on the full longitudinal sequence of a patient's care — every event, in order, in relation to every other event. The architecture is what makes the difference. The difference is what makes the signal appear before the prescription, not after.
A single installation. No per-study procurement. No rebuilding from scratch for each indication.
Each therapy deploys its own pipeline — its own foundation model, trained on that patient population's trajectory. Independent models. Shared infrastructure.
Each pipeline generates multiple forward-looking signals — commercial targeting, patient protection, HEOR evidence — each algorithm optimized for a specific decision.
The value compounds across a portfolio. This is what makes Foresight an infrastructure investment, not a project purchase.
Claims data — prescriptions, utilization, line-of-therapy sequences, adherence signals
EHR data — diagnoses, clinical notes, comorbidities, treatment outcomes
Lab & biomarker data — diagnostic eligibility signals, treatment response markers
Specialty registry data — disease-specific longitudinal data
Unified trajectory graph — every data source connected into a single longitudinal patient view
Forward-looking prediction — models that forecast where cohorts are heading, not just where they've been
Continuous intelligence — population models that update as new data flows in, replacing one-time studies
Claims-first entry point — US engagements begin with your existing claims data and expand from there
Most analytics platforms ask you to buy more data. SHAARPEC Foresight asks what you already have — and makes it work harder.
Built by clinicians and researchers from Harvard Medical School and Mass General Brigham — institutions where clinical pattern recognition and real-world evidence methodology are core disciplines.
Active study with PrecisionAQ — one of the leading pharma commercialization analytics organizations in the US. A national breast cancer cohort of 23,000+ patients, structured from open and closed claims.
Predictive algorithms built and deployed through paid engagements with AstraZeneca, Takeda, Bristol Myers Squibb, Novartis, Novo Nordisk, and Pfizer — on real-world clinical data across multiple therapeutic areas.
CONSIDERING-AF — Trajectory-based atrial fibrillation detection across 330,000 patients. 5.4× improvement vs. standard of care. AUC 0.79–0.84 validated internationally. BMJ Open, 2024.
HaRP — Heart Failure Readmission — Explainable machine learning on longitudinal heart failure trajectories, deployed live in a commercial EHR system. JMIR, 2023.
We start with your indication and your data question — not a generic product demo. If the signal is there, we'll show you what it looks like.
Request a Demo