Technology

How AI Reads Indian Prescriptions (And Why It's Hard)

Indian prescriptions are notoriously hard to read — even for pharmacists. Here's how OCR technology is solving this.

Jeevanee Team5 August 2026 6 min read
6 min listen

The Handwriting Problem

If you have ever tried to read an Indian doctor's prescription, you know the struggle. Abbreviated drug names, mixed Hindi and English, inconsistent formats, and handwriting that would challenge a cryptographer.

Yet this piece of paper is the single most important document in a patient's health journey. Getting it wrong is not an inconvenience. It is a safety risk.

Why Indian Prescriptions Are Uniquely Challenging

Indian prescriptions resist easy digitization for several reasons. There is no standardized format, so every doctor has their own layout, abbreviations, and notation style. Drug names appear in English while instructions sometimes appear in Hindi or regional languages. Unlike the United States and European Union, where electronic prescriptions are common, more than seventy percent of Indian prescriptions are still handwritten. Abbreviations add another layer of complexity. "Tab Met 500 mg BD PC" means "Tablet Metformin 500 milligrams twice daily after food," but only if you know the codes. And a typical chronic care prescription often includes four to seven medications with different schedules.

How OCR Handles This

Modern OCR, or Optical Character Recognition, goes far beyond just reading text from images. Here is the pipeline:

  • Image preprocessing: Adjusting contrast, removing noise, and straightening rotated photos.
  • Layout detection: Understanding which parts are drug names, dosages, frequencies, and instructions.
  • Text recognition: The actual character-by-character reading, trained on Indian handwriting patterns.
  • Medical natural language processing: Understanding that "BD" means twice daily, "PC" means after food, and "Tab" means tablet.
  • Drug name matching: Fuzzy-matching extracted text against a database of Indian pharmaceutical names.
  • The Human-in-the-Loop Principle

    No OCR system is one hundred percent accurate on handwritten prescriptions. That is why Jeevanee follows a strict principle: artificial intelligence suggests, and humans confirm.

    Every extracted prescription goes through a review screen where you can verify each medicine name, adjust dosages, correct frequencies, and add or remove items. Nothing enters your records without explicit approval. This is not a limitation. It is a safety feature.

    Accuracy Rates

    On printed prescriptions, accuracy exceeds ninety-five percent. On typical handwritten prescriptions, accuracy runs between eighty and eighty-five percent. On challenging handwriting, accuracy falls to sixty or seventy percent. In all cases, the human review step catches errors before they matter.

    The goal is not perfection. It is making the process ten times faster than manual entry while keeping safety absolute.

    One place for your family's life

    Health, finances, fitness, and wellbeing — all in one place. Free for families up to 3 members. Set up in under 2 minutes.