TruthNexus
Clinical Decision Support · Medical

CDS that clinicians actually trust.

80%+ of CDS alerts are ignored — because existing systems fire too many false alarms, provide no source attribution, and offer no way to assess how confident an alert is. Skippy Clinical delivers ECE-calibrated recommendations with full evidence tracing, patient-context-adjusted pathways, and FHIR R4 native output at sub-100ms latency.

POST /api/query · 200 OK · 87ms
QUERYwarfarin dosing in elderly with atrial fibrillation
0.856
Confidence
45
Findings
0.07
ECE

Most major guidelines now prefer DOACs over warfarin in patients ≥75 with AF due to lower intracranial hemorrhage risk. If warfarin selected: initiate 2–5 mg/day, INR target 2.0–3.0, check within 3 days.

ACC/AHA 2023FDA LabelsPubMedclinician
80%+
Of CDS alerts ignored by clinicians
<100ms
Response latency at 1,000 RPS
ECE <0.10
Confidence calibration target
12
Medical specialties covered
See It In Action

What a response looks like

Clinical query with reasoning trace, patient-context pathway analysis, and drug-condition evidence ranking — three capabilities, one API.

POST /api/query · application/jsonsimulation
Persona
QUERYwarfarin dosing in elderly with atrial fibrillation
0.856
Confidence · ECE 0.07
45 findings · 87ms
ACC/AHA 2023FDA LabelsPubMed
Answer · clinician

In patients ≥75 with AF, warfarin remains effective but bleeding risk is substantially elevated. Target INR 2.0–3.0. Most major guidelines (ACC/AHA 2023, ESC 2020) now prefer DOACs (apixaban, rivaroxaban) over warfarin in this population due to lower intracranial hemorrhage rates. If warfarin is selected: initiate at 2–5 mg daily, check INR within 3 days, and increase monitoring frequency until stable. CHA₂DS₂-VASc and HAS-BLED scoring before initiation.

Finding Grounding45 findings retrieved — avg confidence 0.856
Confidence CalibrationECE score: 0.07
Source Attribution3 source domains cited
RelatedINR monitoringDOACsCHA₂DS₂-VAScanticoagulation reversal

Simulated output representative of real API responses. ECE-calibrated confidence scores. Findings sourced from ACC/AHA, FDA Labels, ADA, PubMed, and 50+ additional sources.

What It Does

Six capabilities. One API.

POST /api/query

Clinical query

Submit any clinical question and receive a source-attributed, ECE-calibrated answer. Persona-aware responses for clinician, researcher, student, and patient audiences. Returns confidence score, finding count, reasoning trace, and related topics. Target latency: <100ms.

POST /api/clinical/evidence

Drug-condition evidence ranking

Retrieve ranked evidence for a drug-condition pair, filtered by minimum confidence. Each finding returns its confidence score, source, and evidence layer (L1 foundational / L2 supported / L3 clinical). Overall confidence aggregated across the evidence set.

POST /api/clinical/pathway

Patient-context pathway analysis

Analyze a drug's mechanism pathway adjusted for a specific patient — age, conditions, current medications. Returns pathway steps with per-step confidence, flagged drug-drug interactions with severity scores, contraindications, and overall risk (LOW / MODERATE / HIGH).

GET /api/clinical/fhir/{drug}

FHIR R4 MedicationKnowledge

Return a structured FHIR R4 MedicationKnowledge resource for any drug — RxNorm coding, synonyms, intended routes, and confidence-annotated metadata. Ready for Epic CDS Hooks order-select and SMART on FHIR app launch contexts.

POST /api/entity

Entity deep-dive

Retrieve all findings about any clinical entity — drug, condition, mechanism, or biomarker. Findings grouped by type (definition, function, clinical, interaction, contraindication) with per-finding source attribution and confidence.

POST /api/clinical/narrative

Clinical narrative generation

Generate a patient-or-professional clinical narrative from ranked evidence. Audience-aware prose with embedded confidence and source citations — useful for discharge summaries, patient education, and guideline synopses.

Coverage

12 medical specialties. One evidence base.

Guidelines from ACC/AHA, IDSA, NCCN, ADA, USPSTF, and 10+ specialty societies — continuously indexed and attributed. Every recommendation traces back to its source document, version, and effective date.

Cardiology

ACC/AHA heart failure staging, AF anticoagulation, QTc monitoring

Oncology

NCCN treatment protocols, checkpoint inhibitor toxicity, tumor board support

Infectious Disease

IDSA antibiotic selection, resistance patterns, empirical coverage

Endocrinology

ADA diabetes management, thyroid dosing, adrenal insufficiency

Nephrology

eGFR-adjusted dosing, CKD staging, dialysis medication guidance

Neurology

Epilepsy drug interactions, stroke prophylaxis, Parkinson's management

Pulmonology

COPD step therapy, asthma biologics, inhaler technique guidance

Rheumatology

DMARD selection, biologics monitoring, steroid tapering

Psychiatry

Antipsychotic switching, SSRI titration, drug-drug interactions in polypharmacy

Geriatrics

Beers Criteria flagging, fall-risk medications, age-adjusted dosing

Pediatrics

Weight-based dosing, pediatric-specific formulations, off-label guidance

Hematology

Anticoagulation management, transfusion thresholds, bleeding disorder protocols

The alert fatigue problem

Clinicians override 80–95% of drug-allergy and drug-interaction alerts in most hospital systems. The alerts that matter get lost in the noise — and the ones that fire incorrectly train clinicians to dismiss everything.

Skippy Clinical addresses alert fatigue at the source: ECE-calibrated confidence scores let clinicians see exactly how certain a recommendation is, full source attribution lets them verify it, and patient-context awareness reduces false alarms by adjusting for age, renal function, and current medications before an alert fires.

What uncalibrated CDS gets wrong
“A CDS system that says 'high confidence' on 90% of alerts — regardless of actual accuracy — provides no useful signal. Calibration means the confidence score predicts accuracy. ECE < 0.10 means a 0.85 score is right 85% ± 10% of the time.”

Expected Calibration Error is reported on every Skippy Clinical response alongside the confidence score. Clinicians and system administrators can audit calibration performance over time — and trust is earned through transparent measurement, not assertions.

Who It's For

EHR vendors, health systems, and specialty clinics

EHR Vendors
Add regulatory-grade CDS content to Epic, Cerner, or athenahealth without building an evidence layer from scratch. CDS Hooks and SMART on FHIR integration patterns included.
Health Systems
Replace uncalibrated alert engines with evidence-sourced decision support that clinicians can verify. Source attribution reduces alert override rates by giving clinicians a reason to act.
Specialty Clinics
Cardiology, oncology, and infectious disease recommendations grounded in specialty-specific guidelines — ACC/AHA, NCCN, IDSA — updated in real time as guidelines change.
Academic Medical Centers
Tamper-evident audit trails with full reasoning traces for accreditation, outcomes research, and Joint Commission medication safety documentation.
Regulatory

Built for regulated clinical AI deployment

ECE-calibrated confidence scores align with FDA SaMD AI/ML guidance requirements for performance transparency. Source attribution and tamper-evident audit trails satisfy Joint Commission medication safety standards and ONC HTI-1 predictive decision support documentation requirements.

FHIR R4 native output aligns with CMS-0057-F prior authorization requirements. SMART on FHIR launch context supported for Epic and Cerner. HIPAA-ready with BAA available. FDA 510(k) clearance pathway underway (Class II SaMD).

SOC 2 Type II certification targeted Q4 2026. Full reasoning trace and finding provenance retained per response for audit and accreditation purposes.

HIPAA ReadyONC HTI-1FHIR R4SMART on FHIRFDA 510(k) In Progress
Clinical Use

Skippy Clinical is a decision-support tool for qualified healthcare professionals. It does not replace physician judgment, pharmacist review, or institutional clinical protocols. All recommendations should be confirmed against current clinical guidelines, patient-specific factors, and applicable institutional policies before clinical action. FDA 510(k) clearance in progress; not yet cleared as a medical device. HIPAA-ready; BAA available.

See Clinical in your EHR workflow

We work with EHR vendors, health systems, and specialty clinics. Let's talk about your clinical decision support problem.