🩺 Clinical Research & Longitudinal Care

Three years of labs, one longitudinal view

Biomarker-intervention tracking. Contradiction detection. Pattern discovery across cohorts.

Patient longitudinal profiles connecting biomarkers to interventions across years. Schema induction discovers treatment response patterns across cohorts. Pensiv turns fragmented clinical data into compounding clinical insight.

What you're ingesting

Clinical data sources

Lab results

Blood panels
Biomarker assays
Metabolic profiles
Genetic testing results

Clinical notes

Physician notes
Treatment plans
Progress notes
Discharge summaries

Imaging & diagnostics

Imaging reports (MRI, CT, X-ray)
Radiology findings
Pathology reports
Biopsy results

Medication history

Prescriptions
Dosage changes
Adverse events
Medication interactions

Published research

Clinical trial data
Peer-reviewed studies
Treatment guidelines
Meta-analyses

Patient-generated data

Wearable device exports
Symptom logs
Patient-reported outcomes
Adherence tracking

What the vault builds

Longitudinal profiles and treatment patterns

Patient longitudinal profiles connecting biomarkers to interventions across years

The vault tracks every lab, every intervention, every outcome over time. When HbA1c trends upward 6 months after medication change, the connection is explicit and traceable.

Schema induction discovers treatment response patterns across cohorts

After 50+ patients with similar baseline profiles, patterns emerge: "Patients with [biomarker profile X] respond to [intervention Y] within 8-12 weeks." No manual coding required.

Contradiction detection flags when new lab results conflict with treatment hypothesis

The system doesn't just store — it thinks. When a biomarker moves in the opposite direction from what the treatment plan predicts, Pensiv surfaces the contradiction for review.

Temporal filtering isolates changes within specific windows

"What changed between month 6 and month 12?" The vault knows event time (when it happened) vs ingestion time (when you learned). Query the patient's state at any moment.

What you ask it

Example queries for clinical researchers

"Across this patient's last 3 years of labs, which biomarkers are trending in a direction that contradicts the current treatment plan?"

What Pensiv returns Contradiction detected: HbA1c rising +0.8% over 6mo despite medication intensification (expected: -0.5 to -1.0%). Insulin resistance markers (HOMA-IR) also worsening. Last intervention (metformin dose increase, Mar 2026) shows no effect. Schema suggests GLP-1 agonist addition based on 12 similar patient profiles.

"Find patients with similar baseline biomarker profiles who responded to [intervention X]"

What Pensiv returns 23 patients with comparable baseline (HbA1c 7.8-8.4%, BMI 28-32, age 45-60). 18/23 responded to GLP-1 within 12 weeks (HbA1c -0.9 to -1.4%). 5 non-responders shared elevated inflammatory markers (CRP >3.0). Schema family: "GLP-1 response pattern in metabolic syndrome cohort."

"What published research supports or contradicts the treatment trajectory we're seeing?"

What Pensiv returns Supporting: 4 RCTs (2022-2025) show GLP-1 efficacy in similar cohort. Contradicting: 1 meta-analysis (2024) found reduced efficacy in patients >55 with baseline CRP >4.0 — matches your patient profile. Recommendation: Consider anti-inflammatory co-treatment per Smith et al. (2025).

"What intervention changes preceded the biomarker shift we're seeing?"

What Pensiv returns Temporal sequence: Statin dose increase (Jan 15) → 8 weeks → LDL dropped 22% (expected) + ALT/AST elevation (unexpected, possible statin hepatotoxicity). No other medication changes. Contradiction: treatment hypothesis predicted LDL improvement only. Schema: "statin-induced enzyme elevation pattern."

Compliance & security

HIPAA-ready, on-prem deployment

HIPAA BAA

Business Associate Agreement. Full audit trails, encryption at rest and in transit, access controls, breach notification.

On-prem deployment

Run on your infrastructure. PHI never leaves your network. One database to back up, full control.

De-identification support

Entity extraction automatically flags PII/PHI. Safe Harbor and Expert Determination workflows for research use.

Audit trails

Every query logged. Full provenance from clinical note → extracted entity → retrieval → synthesis. Required for IRB and regulatory filings.

Use cases

Who this is for

Clinical researchers

Longitudinal cohort studies. Pattern discovery across treatment arms. Cross-study meta-analysis with full provenance for publications.

Precision medicine teams

Match patient profiles to treatment response patterns. Contradiction detection when biomarkers diverge from expected trajectories.

Concierge & longevity medicine

Multi-year patient profiles. Biomarker tracking across interventions (diet, supplements, medications). Personalized pattern discovery.

Health-tech platforms

Embed Pensiv as the memory layer for AI-powered clinical decision support. MCP integration, HIPAA-ready, full audit trails.

Ready to turn fragmented data into longitudinal insight?

Patient longitudinal profiles. Treatment response pattern discovery. Contradiction detection. HIPAA-ready on-prem deployment.