Personalized Prescription Generator Development

Project Title

Development of a Multi-Clinic Doctor Prescription Generator with Personalized Handwriting and Voice-to-Prescription

1. Project Overview

We are looking for an experienced full-stack developer or small software development team to develop a secure, responsive, multi-tenant web application for doctors and clinics.

The application will allow doctors to register patients, select medicines, enter clinical instructions by typing or speaking, and generate professional A4 prescriptions in a personalized blue-ink handwriting style.

Each doctor must have an independently managed handwriting profile created using uploaded photographs or scanned pages of their handwriting.

The application must support desktop computers, laptops, tablets and Android mobile browsers.

Prescription language: English only for Version 2.0.

2. Main Features

A. Multi-Clinic and Doctor Management

- Multiple clinics, doctors and authorized staff.
- Secure registration and login.
- Role-based access control.
- Independent doctor profiles and clinic settings.
- Strict separation of each clinic's patient and prescription data.

B. Patient Management

- Patient registration and unique patient IDs.
- Patient details, allergies and relevant medical history.
- Visit history and previous prescriptions.
- Search and duplicate-patient detection.

C. Medicine Catalogue and Prescription Editor

- Searchable medicine catalogue.
- Brand and generic medicine names.
- Strength, formulation and active ingredient.
- Dose, frequency, route, duration, quantity and instructions.
- Support for manually entered medicines.
- Editable clinical notes, diagnosis, advice and follow-up.

The application must never independently prescribe medicines, infer missing doses or silently change clinical instructions.

D. Personalized Doctor Handwriting

- Upload JPG, PNG and scanned PDF handwriting samples.
- Image cropping, rotation and basic cleanup.
- Individual handwriting profiles for each doctor.
- Adjustable letter size, spacing, slant and appearance.
- Blue-ink handwriting-style rendering.
- Preview and comparison of different styles.
- Versioned profiles with rollback support.

Initially, a licensed handwriting font with configurable styling may be used. A personalized AI handwriting-generation model may be proposed as a separate phase.

The developer must clearly distinguish font-based styling from genuine sample-based personalization. No claim of exact handwriting reproduction will be accepted without successful validation.

E. Voice-to-Prescription

- Start, pause, resume and stop voice dictation.
- Support compatible desktop and Android microphones.
- English speech recognition.
- Dictation into selected clinical fields.
- Editable transcription before approval.
- Identification of uncertain terms and potential medicine-name, dosage and negation errors.
- Manual text-entry fallback.

All recognized clinical content must be reviewed and explicitly approved by the doctor. Voice recognition must never independently authorize or finalize a prescription.

Raw audio must not be retained by default.

F. Prescription PDF and Printing

- Clinic letterhead and logo.
- Doctor name, qualifications, specialty and registration number.
- Patient details and prescription date.
- Clinical notes and medicine instructions.
- Doctor-authorized signature and stamp.
- A4 colour PDF generation.
- Browser print preview and reprinting.
- Readable text, consistent page breaks and reliable layout.
- Preservation of the exact approved prescription version.

G. Validation, Security and Audit

- Canonical storage of approved clinical text.
- OCR-based comparison of rendered text against the approved record.
- Flagging of medicine names, numbers, decimal points, units and dosage mismatches.
- Readable typed fallback for unreliable handwriting output.
- Explicit doctor approval before finalization.
- Audit logs for changes, profile updates and approvals.
- Encryption in transit and at rest.
- Secure backups and tested recovery.
- Restricted access to handwriting samples, patient records, signatures and stamps.

OCR must not be treated as a guarantee of clinical correctness.

3. Suggested Technology Stack

Preferred technologies:

- Frontend: React and TypeScript.
- Backend: Python, Django and Django REST Framework.
- Database: PostgreSQL.
- Background jobs: Celery and Redis, or equivalent.
- Image processing: OpenCV or equivalent.
- OCR: A suitable validated recognition engine.
- PDF generation: HTML/CSS-to-PDF or an equivalent reliable solution.
- Storage: Private encrypted object storage.
- Deployment: Secure HTTPS-enabled cloud infrastructure.

Developers may suggest alternatives with a clear explanation of cost, security, scalability and maintenance implications.

4. Project Phases

Phase 1 — Core MVP

- Authentication and clinic management.
- Patient and medicine management.
- Prescription editor.
- Handwriting-style rendering.
- A4 PDF generation and printing.
- Doctor approval and audit logging.

Phase 2 — Personalization

- Handwriting sample upload and preprocessing.
- Individual profiles and configurable styles.
- Profile versioning and preview.
- Text-integrity validation.

Phase 3 — Voice Input

- Speech recognition and editable transcription.
- Field-specific dictation.
- Recognition-error handling.
- Doctor confirmation and audit trail.

Phase 4 — Advanced Handwriting AI

- Evaluate sample-based handwriting generation.
- Develop an isolated model service if feasible.
- Test unseen text, medicine names, numbers and dosages.
- Implement validation, fallback and rollback.
- Demonstrate performance before production use.

Each phase must have independently testable deliverables and acceptance criteria.

5. Required Deliverables

The selected freelancer must provide:

1. Complete working source code.
2. Responsive web application.
3. Database schema and migration files.
4. API documentation.
5. Deployment instructions and configuration templates.
6. Working development and production deployment.
7. PDF and printing functionality.
8. Voice-dictation functionality.
9. Handwriting profile and sample-management functionality.
10. Automated tests for core workflows and tenant isolation.
11. Security and data-protection documentation.
12. User and administrator documentation.
13. Test reports covering handwriting and clinical text integrity.
14. Source-code repository and handover.
15. A demonstration of all agreed acceptance criteria.

All agreed intellectual property, source code and project assets must be transferred to the client under the contract. Third-party software and font licences must be documented.

6. Required Developer Experience

Applicants should have demonstrated experience in:

- Full-stack web application development.
- Python/Django or an equivalent backend.
- React or a comparable frontend framework.
- Relational database design.
- Secure multi-tenant applications.
- PDF generation and document rendering.
- OCR or image processing.
- Speech-to-text integration.
- AI/ML model integration, where applicable.
- Secure handling of confidential information.

Experience with healthcare software is an advantage.

7. Proposal Requirements

Please include:

1. Relevant projects and portfolio links.
2. Your proposed technical architecture.
3. Recommended handwriting approach and its limitations.
4. Proposed speech-recognition service and privacy implications.
5. Fixed-price estimate for each phase.
6. Estimated timeline for each phase.
7. Hosting and recurring operating costs.
8. Testing, security and handover plan.
9. Post-delivery bug-fix and maintenance terms.
10. Any assumptions, exclusions or third-party licensing costs.

Please identify which features are included in the initial MVP and which require separate development.

8. Acceptance Criteria

The project will be accepted only when:

- Multiple clinics operate with verified tenant isolation.
- Doctors can create patients and prescriptions.
- Doctors can upload and manage individual handwriting samples.
- Voice input produces editable clinical text.
- Unapproved dictated content cannot appear in a finalized prescription.
- The application preserves the exact approved clinical wording.
- Unreadable or mismatched output is flagged for review.
- A4 colour PDFs print correctly.
- Only authorized users can apply signature and stamp assets.
- Prescription approvals and changes are auditable.
- The application passes the agreed security, functional and recovery tests.
- Complete source code and deployment documentation are handed over.

9. Budget and Payment

Please quote a fixed price for each phase rather than one unexplained amount for the entire project.

The proposal must separately identify:

- Development costs.
- Third-party licences and APIs.
- Cloud hosting and storage.
- Speech-recognition usage.
- Optional AI training and GPU costs.
- Maintenance and support.

Payments will be linked to demonstrated milestones and acceptance tests.

10. Important Project Conditions

- Clinical decisions remain entirely under the doctor's control.
- No unapproved AI-generated clinical content may be finalized.
- Patient data and handwriting samples must not be used for unrelated model training.
- All external services and data-processing arrangements must be disclosed.
- Advanced handwriting AI must be demonstrated and validated before it is considered production-ready.
- No production patient data will be used for initial demonstrations without appropriate authorization and safeguards.

Please submit a phased technical and commercial proposal with a working demonstration plan.