Skip to content

Making prescriptions and lab reports understandable, in plain Bangla.

An AI system inside the Health 360 app that reads a photo of a prescription or lab report and explains it to patients in plain Bangla, and says so clearly when it is not sure.

From a photo to an explanation
  1. Photo of a prescription or report01

    Taken by the patient in the app

  2. Image quality check02

    Pre-processing before anything is read

  3. Information extracted03

    Vision language model reads the document

  4. Confidence scored04

    Every result carries a reliability indicator

  5. Explained in plain Bangla05

    With a clear prompt to consult a doctor

  6. Uncertain? Show a pharmacist06

    The system says so instead of guessing

Client
Health 360
Market
Bangladesh
Users
Patients and pharmacies
Status
MVP delivered

01Context

Most patients cannot read their own prescription.

In Bangladesh, prescriptions are usually handwritten, medicine names are in English, and doses are abbreviated. Many patients cannot read them.

The consequences are ordinary and serious: medicine taken at the wrong time, in the wrong amount, or a course left unfinished. Lab reports create the same problem. Patients see numbers but cannot tell which ones are normal.

02Constraint

The constraint

Health 360 already had a health app, but nothing in it helped a patient understand the documents their care depended on.

03Understanding

What we understood

Reading the document was only half of the problem. In healthcare, a confident wrong answer is worse than no answer. The system had to explain clearly when it was sure, and say plainly when it was not.

04The system

What the system produces

From a prescriptionPer medicine
Medicine name
Strength
Daily dose
Time of day
Before or after food
Duration

Alongside each medicine: why it is commonly prescribed, common side effects and precautions.

Reliability
Illustrative structure, not a product screenshot
From a lab report
ParameterPatient's valueReport's own reference rangeNormal, high or low
Normal
High
Normal
Low

Each parameter is compared with the reference range printed on that report, then explained in simple terms.

Illustrative structure, not a product screenshot

05Safety design

Designed for uncertainty, not just accuracy.

Safety was part of the design from the start, not a disclaimer added at the end.

  • 01

    A reliability indicator on every result

    Patients can see how confident the system is about each extracted value.

  • 02

    Honest when unsure

    When the system is uncertain, it says so clearly and asks the patient to show the document to a pharmacist.

  • 03

    Doctor's advice always comes first

    Every report ends with guidance to consult a doctor.

06Value

Built for two kinds of users

Patients

  • See their medicine routine clearly
  • Save reports to keep track
  • Show saved reports to their doctor later

Pharmacies

  • Less time spent deciphering prescriptions at the counter
  • Designed to reduce the risk of dispensing errors where no trained pharmacist is available

07Technology

Extraction
Vision language model reading information directly from photos
Image pipeline
Quality checks and pre-processing before extraction
Reliability
Confidence scoring for every extracted value
Output
Plain-Bangla report generation
Integration
Integrated into the Health 360 Flutter app

08Status

Where it stands

The MVP has been delivered to Health 360. We will publish outcomes here only when they have been measured and approved by the client.

09Reflection

What it taught us

In high-stakes domains, the most important design decision is what the system does when it is not sure. Telling a patient to ask a pharmacist is a feature, not a failure.

Have a problem worth solving properly?

Describe the problem in your own words. We will tell you honestly whether we can help, and how we would approach it.