Review Verification Methodology
In standard search engines and consumer directories, up to 40% of healthcare reviews are astroturfed, incentivized, or fabricated. BioTB is built on a non-negotiable premise: every single patient review must be clinically authenticated against real medical documentation.
Layer 1: Multimodal OCR & LLM Prescription Parsing
When a patient submits a review, they upload an image or PDF of their OPD consultation receipt, discharge summary, or doctor's prescription. Our ingestion pipeline executes an automated 4-step forensic scan:
Before any document is processed by LLM analyzers or human moderation eyes, a local computer vision model detects and permanently blackens patient name, mobile numbers, Aadhaar/government IDs, and home addresses in memory.
High-resolution OCR extracts the doctor's printed/stamped name, medical council registration number (e.g. DMC-14283), clinic letterhead, and date of consultation.
Our clinical NLP model analyzes prescribed medicines (e.g. Atorvastatin, Metformin, Inhaled Budesonide) and cross-checks them against the patient's reported symptoms to guarantee review relevance.
Every uploaded document is converted into a normalized perceptual hash (pHash) and SHA-256 fingerprint. Reused prescriptions across different user accounts are instantly flagged as fraud.
Layer 2: Statutory Medical Council Registry Cross-Validation
BioTB maintains automated and manual reconciliation bridges with statutory medical registers:
Layer 3: On-Ground Field Auditing in Delhi NCR
Unlike purely algorithmic platforms, BioTB deploys trained healthcare field auditors across Delhi, Gurgaon, Noida, and Faridabad:
Auditors verify consultation timings, actual consultation fees, waiting room hygiene, and accessibility, ensuring listing parameters match reality.
With patient consent at OPD exits, field teams collect verified audio/written feedback that serves as ground-truth anchor data against digital reviews.
4. The Mathematical BioTB TrustScore™ Engine
The BioTB TrustScore™ (scaled 0 to 10) is a weighted deterministic composite calculated via:
5. The Zero Pay-to-Rank Guarantee
Traditional directories operate on "sponsored listing" models where the highest paying hospital appears at the top. BioTB explicitly repudiates this practice:
- Listing rank is determined strictly by algorithmic TrustScore, clinical proximity, and sub-specialty match.
- There is no "pay to hide bad reviews" button. If a patient submits a negative review accompanied by an authentic prescription, it stays published.
- Providers have the right to submit clinical rebuttals or document corrective resolution, which is visibly published alongside the patient review.