Medical Prescription - Intelligent Formatting & Digitization using AI
Medical prescriptions contain the necessary information to acquire the necessary medicines for recovery and a healthy life. Important information on prescriptions is not just a mere list of medications; rather, they contain detailed information on how the medication is to be dispensed, what brand (or lab) the medication came from, preferred duration of medication use, patient history, etc.
But just like registration forms, they must be properly filed and secured to prevent confusion, misuse and erroneous prescriptions. There's a high level of ethical and legal precedence regarding the usage and the continuation of these prescriptions. They have to be safeguarded and used well for their respective, proper purposes.
New Delhi, India
Client
Healthcare Technology Solution Provider
CLIENT
Our client is a healthcare focused IT services and consulting company in India. Embebo.com offers hospitals, clinics and healthcare facilities the best customer relationship and records management. Embebo offers modules & tools that can organize your medical notes, records, and opinions with seamless integration, efficient control, and enhanced insights.
We were tasked by Embebo to create an AI-based system that could intelligently format and digitize medical prescriptions produced by doctors and hospitals.
CHALLENGE
We encounter a set of three challenges before implementing the concept that we have created, they are listed below:
-
Archiving of prescriptions - Prescriptions are legal-binding documents, hence their safety and content security is apex. No external nor internal threats should affect them and they should be only edited with proper permissions.
-
Extracting text from medical writing style - It is common knowledge that doctors use a different script to write prescriptions for their own reference. Extracting these types of scripts/handwriting is hard even by modern NLP standards, and they would need a cloud-backing for efficient functionality.
-
Prescription Pattern Identification - Prescriptions have varying sets of patterns but they universally contain important values such as Medicine Name, Dosage, and etc. To create a rule-set that fits to this criteria is important.
SOLUTION
The solution we have devised for this problem and its set of challenges is to “dynamically format” and digitize the prescriptions utilizing AI methodology.
-
We use our own in-house Medical OCR (optical character recognition) program to systematically change and append changes to prescriptions.
-
In conjunction with AWS’ Textract advanced text/handwriting detection algorithm, we are able to assess handwritten prescription's accuracy at scale. This is done by extracting the basic components of any prescription (for example, a chief complaint and diagnosis), which are called rules.
-
These “rules” provide the backbone for filtering the contents of the prescription to specification. With these rules, we can safeguard the contents and the critical fields of every prescription. There's no systemic nor legal issues with this, proving the rule-based method to be most viable option.
Identified text bounding boxes
Editable digital response
RESULT
-
With our ground-breaking prescription digitization technology, it would be very easy for doctors and patients alike to refer to their prescriptions.
-
There would be no error in dosage nor the medication prescribed; preventing dangerous side-effects from occurring. Implementing this module within their existing customer management system will save countless lives and earn a name for Embebo in their market for catering to safety and security by intelligent methods.