Skip to content

Vision & Strategy

Executive Summary

Entheory.AI is building India's first AI-powered longitudinal oncology record system that unifies fragmented cancer care data across EMRs, labs, imaging, and genomics platforms. Our mission is to provide oncologists with a single, comprehensive patient view that enables better treatment decisions while respecting India's unique healthcare infrastructure and linguistic diversity.

1. Vision (3-5 Years)

Our Long-Term Vision

To become the standard longitudinal oncology data platform across tier-1 and tier-2 hospitals in India, enabling:

  • Data Continuity: Every cancer patient's journey is captured from diagnosis through treatment, regardless of which healthcare facility they visit
  • AI-Assisted Decision Support: Clinicians receive evidence-based treatment recommendations powered by longitudinal patient data and medical literature
  • Research Acceleration: Anonymized, structured oncology data enables Indian cancer research and clinical trials
  • Patient Empowerment: Patients own and access their complete cancer records via ABHA (Ayushman Bharat Health Account)

Success Metrics (Year 3-5)

Metric Target
Hospitals Deployed 50+ (Tier 1 & 2 hospitals)
Patient Records 100,000+ oncology patients
Data Completeness >90% (all modalities present)
Clinician Adoption >80% daily active usage
Research Datasets 5+ published studies using platform data

2. Mission (Next 12-24 Months)

What We're Solving Now

Deploy a trustworthy, hospital-grade longitudinal oncology record at 3-5 pilot hospitals, proving:

  1. Technical Feasibility: Can integrate with existing EMRs, LIS, PACS without disrupting workflows
  2. Clinical Value: Oncologists save time and make better decisions with unified patient view
  3. Data Quality: OCR/ASR handles English + Hindi medical documents with >85% accuracy
  4. Regulatory Compliance: Meets ABDM, DISHA, and hospital security/privacy requirements

MVP Scope

  • Single hospital deployment with full data integration (HL7, FHIR, JSON feeds)
  • Longitudinal patient timeline (labs, imaging, pathology, genomics, therapy)
  • Bilingual support (English + Hindi for OCR/ASR)
  • FHIR R4 export for interoperability
  • Basic rule-based alerts (critical labs, treatment delays)

3. Problem Statement

The Challenge

Oncology care in India suffers from severe data fragmentation:

3.1 Fragmented Systems

EMR/HIMS          LIS (Labs)        PACS (Imaging)      Genomics Lab
    ↓                  ↓                  ↓                   ↓
Different vendors, no interoperability, proprietary formats
    ↓                  ↓                  ↓                   ↓
Oncologist must check 4-6 separate systems to see one patient

Impact: - 15-20 minutes wasted per patient gathering data from multiple systems - Missing data: 30-40% of relevant historical data not accessible during consultation - Treatment delays: Oncologists postpone decisions waiting for reports from other systems

3.2 Lack of Longitudinal View

  • No timeline: Can't easily see patient's cancer journey over months/years
  • No trends: Lab values, imaging measurements not visualized longitudinally
  • No context: New test results reviewed in isolation without historical comparison

3.3 India-Specific Constraints

Challenge Impact
Multilingual documentation Medical notes in mix of English, Hindi, regional languages
Unstructured data 60%+ data in PDFs, handwritten notes, audio recordings
Infrastructure gaps Small hospitals lack full EMRs, use paper + basic systems
Cost sensitivity Hospitals won't replace existing EMRs
Regulatory complexity ABDM/NDHM compliance required, Digital Personal Data Protection Act 2023

3.4 Missed Opportunities

  • No AI/ML insights possible without longitudinal, structured data
  • Research barriers: Retrospective studies impossible due to data silos
  • Quality gaps: No way to track treatment outcomes, adverse events systematically

4. Our Solution: Entheory.AI Platform

4.1 Core Principles

  1. Layer, Don't Replace
  2. Integrate on top of existing EMRs, PACS, LIS
  3. No rip-and-replace; hospitals keep current systems

  4. Lightweight deployment (weeks, not months)

  5. India-First Design

  6. Bilingual (English + Hindi), expandable to regional languages
  7. OCR/ASR for unstructured data (PDFs, audio)

  8. ABDM-aligned (ABHA IDs, FHIR R4)

  9. Trust & Transparency

  10. Provenance tracking: Every data point shows source, timestamp
  11. No black-box AI: Rule-based alerts only initially; LLM-based features clearly marked
  12. Clinician control: Physicians can audit all automated processes

4.2 Architecture Overview

┌─────────────────────────────────────────────────────────────┐
│                    Physician Interface (UI)                  │
│   Timeline | Labs | Imaging | Genomics | Therapy | Notes   │
└─────────────────────────────────────────────────────────────┘
                              ↑
                      ┌───────┴────────┐
                      │   REST APIs    │
                      │  (JSON / FHIR) │
                      └───────┬────────┘
                              ↓
┌─────────────────────────────────────────────────────────────┐
│        Canonical Patient Bundle (Longitudinal Record)        │
│   Single source of truth per patient (JSON + FHIR R4)      │
└─────────────────────────────────────────────────────────────┘
                              ↑
            ┌─────────────────┼─────────────────┐
            ↓                 ↓                  ↓
    ┌──────────────┐  ┌──────────────┐  ┌──────────────┐
    │ Ingestion    │  │ Processing   │  │ Integration  │
    │              │  │              │  │              │
    │ • HL7 v2     │  │ • OCR (Eng+Hi)│  │ • FHIR Export│
    │ • FHIR R4    │  │ • ASR (Eng+Hi)│  │ • ABDM Sync  │
    │ • JSON Feeds │  │ • Validation │  │ • PHR App    │
    └──────────────┘  └──────────────┘  └──────────────┘
            ↑                 ↑                  ↑
     ┌──────┴────────┬───────┴────────┬─────────┴─────┐
     ↓               ↓                ↓               ↓
┌─────────┐  ┌──────────────┐  ┌──────────┐  ┌──────────────┐
│ EMR/HIMS│  │ LIS (Labs)   │  │   PACS   │  │ Genomics Lab │
└─────────┘  └──────────────┘  └──────────┘  └──────────────┘

4.3 Key Differentiators

Feature Traditional EMR Entheory.AI
Scope Hospital-specific Cross-facility longitudinal
Integration Monolithic Modular, pluggable adapters
Language English only English + Hindi + extensible
Unstructured Data Manual entry Automated OCR/ASR
Interoperability Proprietary FHIR R4, ABDM-compliant
Deployment 6-12 months 2-4 weeks
Cost ₹50L-2Cr capex SaaS, ₹500-2000/patient/year

5. Strategic Objectives

Objective 1: Build Trust with Clinicians

Goal: Oncologists use Entheory.AI for >50% of patient consultations within 6 months

Tactics: - Data accuracy: >95% OCR accuracy, validated against source systems - Speed: Patient record loads in <2 seconds - Reliability: 99.5% uptime, no data loss - Transparency: Clear provenance for all data points - Clinical validation: Pilot with 5-10 trusted oncologists, iterate based on feedback

Metrics: - Daily Active Users (DAU) among physicians - Time saved per consultation (target: 5-10 minutes) - User satisfaction scores (NPS >50)

Objective 2: Prove Integration Feasibility

Goal: Successfully integrate with 3+ heterogeneous hospital systems (EMR, LIS, PACS brands)

Tactics: - Standard protocols: HL7 v2, FHIR R4 adapters - Fallback: JSON file drop for legacy systems - Minimal IT burden: Deployment in <2 weeks per hospital - Error handling: DLQ and alerting for failed ingestions

Metrics: - Integration success rate (target: >95% uptime) - Mean time to integration (target: <2 weeks) - Data completeness (target: >80% of available data ingested)

Objective 3: Demonstrate Indian Language Support

Goal: Hindi OCR/ASR accuracy >80%, enabling use in bilingual hospitals

Tactics: - Model selection: Tesseract (OCR) + Whisper (ASR) with Hindi packs - Benchmarking: Test on 100+ real Hindi medical documents from pilot hospitals - Fallback: Manual review queue for low-confidence extractions - User controls: Allow physicians to flag incorrect OCR/ASR for retraining

Metrics: - Hindi OCR accuracy (target: >80% on scanned documents) - Hindi ASR word error rate (target: <20%) - Bilingual document processing volume (target: 30% of uploads in Hindi)

Objective 4: Align with ABDM/NDHM Ecosystem

Goal: ABDM-compliant by MVP completion; ready for PHR app integration

Tactics: - ABHA ID normalization: All patients mapped to ABHA IDs - FHIR R4 compliance: Validated bundles for Patient, Observation, DiagnosticReport, etc. - Consent management: Prepare for ABDM consent framework (future) - HIP/HIU registration: (Post-MVP but architecturally ready)

Metrics: - 100% of patients with ABHA IDs (or fallback local IDs) - FHIR bundle validation: 0 critical errors - ABDM compliance checklist: 100% complete

6. Non-Goals (Current Phase)

To maintain focus, we explicitly exclude the following from MVP:

6.1 Full-Fledged EMR Replacement

  • Why: Hospitals already invested in EMRs; rip-and-replace is too disruptive and expensive
  • Alternative: We integrate via HL7/FHIR; future may offer lightweight EMR for clinics without existing systems

6.2 Global Multilingual Support

  • Why: Resource constraints; focus on India market first
  • Current: English + Hindi only
  • Future: Add regional languages (Tamil, Telugu, Kannada) based on hospital demand

6.3 Advanced Clinical Decision Support (CDS)

  • Why: Requires longitudinal data volume we don't have yet; regulatory approval needed for AI-based recommendations
  • Current: Basic rule-based alerts (critical labs, treatment delays)
  • Future: LLM-powered insights, guideline adherence checks (with clear disclaimers)

6.4 Full Financial / Billing Workflows

  • Why: Out of core competency; many specialized billing systems exist
  • Current: Display insurance type, basic cost info where available
  • Future: May integrate with billing systems for cost transparency

6.5 Patient-Facing Mobile App (MVP)

  • Why: Focus on physician workflows first; patient app requires separate UX and regulatory considerations
  • Current: Physician-facing web UI only
  • Future: PHR (Personal Health Record) app for patients to access their records

7. Go-to-Market Strategy

Phase 1: Pilot (Months 1-6)

Target: 1-2 hospitals, 500-1000 patients

Activities: - Deploy at 1 tier-1 oncology center and 1 tier-2 hospital - Validate technical integration, clinical workflows - Gather feedback, iterate on UI/UX - Establish baseline metrics (time saved, data completeness)

Success Criteria: - >70% clinician adoption - >85% data completeness across modalities - <5 critical bugs per month

Phase 2: Early Adopters (Months 7-12)

Target: 5-10 hospitals, 5000-10000 patients

Activities: - Expand to hospitals in different cities (geographic diversity) - Test different EMR/LIS/PACS vendor combinations - Build case studies from pilot hospitals - Develop training materials, onboarding playbooks

Success Criteria: - Repeat deployment in <2 weeks per hospital - Customer NPS >40 - 1-2 published case studies showing clinical impact

Phase 3: Scale (Months 13-24)

Target: 30+ hospitals, 50,000+ patients

Activities: - Scale sales and deployment team - Launch partner ecosystem (EMR vendors, PACS vendors, labs) - Develop self-service deployment for smaller clinics - Explore research partnerships for anonymized datasets

Success Criteria: - Revenue: ₹2-5 Cr ARR - Market share: 10-15% of tier-1 oncology centers - Strategic partnerships: 2-3 EMR/PACS vendors

8. Competitive Landscape

8.1 Current Alternatives

Solution Strength Weakness vs. Entheory.AI
Manual EMR navigation Free (existing system) Time-consuming, no longitudinal view, no AI
Custom hospital integrations Tailored to hospital Expensive (₹1Cr+), 12+ month implementation
International oncology EMRs Feature-rich No India-specific workflows, no Hindi support, ₹₹₹
ABDM PHR apps Patient-controlled Physician UI lacking, no clinical workflows

8.2 Our Competitive Edge

  1. India-first: Only solution with Hindi OCR/ASR out of the box
  2. Fast deployment: Weeks vs. months/years
  3. Non-disruptive: Layer on existing systems, don't replace
  4. Oncology-specialized: Deep domain knowledge, not generic EMR
  5. ABDM-native: Built for Indian healthcare ecosystem from ground up

9. Risks & Mitigation

Risk Impact Probability Mitigation
Hospital IT resistance High Medium Executive sponsorship, minimize IT burden (cloud-hosted option)
Data quality issues High High Rigorous validation, manual review workflows for low-confidence OCR/ASR
Regulatory changes (DPDP Act) Medium Medium Legal counsel, stay aligned with ABDM updates
Hindi OCR accuracy <80% Medium Medium Extensive testing, manual fallback, model fine-tuning
Slow clinician adoption High Low User-centric design, feedback loops, training programs

10. Key Performance Indicators (KPIs)

Technical KPIs

  • Integration Uptime: >99.5%
  • Data Completeness: >80% of available modalities ingested
  • OCR/ASR Accuracy: English >90%, Hindi >80%
  • API Response Time: <2s (p95)

Clinical KPIs

  • Time Saved: 5-10 minutes per patient consultation
  • Data Accessibility: <3 clicks to any patient data point
  • Clinician Satisfaction: NPS >40

Business KPIs

  • Hospitals Deployed: 3-5 (Year 1), 30+ (Year 2)
  • Revenue: ₹50L (Year 1), ₹2-5Cr (Year 2)
  • Patient Records: 5K (Year 1), 50K (Year 2)

11. Next Steps

Immediate (Next 30 Days)

  • [ ] Finalize MVP scope and technical architecture
  • [ ] Select pilot hospital partner
  • [ ] Set up development environment and CI/CD
  • [ ] Begin EMR/LIS/PACS integration development

Short-Term (Next 90 Days)

  • [ ] Complete MVP development
  • [ ] Deploy at first pilot hospital
  • [ ] Conduct user testing with 5-10 oncologists
  • [ ] Iterate based on feedback
  • [ ] Prepare case study materials

Medium-Term (6-12 Months)

  • [ ] Expand to 5 hospitals
  • [ ] Prove Hindi OCR/ASR at scale
  • [ ] Develop sales and marketing materials
  • [ ] Seek seed/Series A funding
  • [ ] Build partnerships with EMR/PACS vendors

Document Owner: Product & Strategy Team
Last Updated: 2024-12-03
Next Review: End of Q1 2025 (post-pilot evaluation)