Medical IDP (Intelligent Document Processing) 
Platform

Documentation is key for the proper functioning for any field. For a critical domains like the medical and healthcare industry; documentation, in a literal sense, deals with life and death. Medical Documentation provides the industry with a definitive chronicle of its fields of activity which also doubles as a reference material for future diagnostics and development.

Medical documents are naturally data-intensive; they contain a plethora of cases and information that are intertwined with various disciplines. They contain both printed, scanned and handwritten matter, the latter being difficult to extract due to the stenographic nature of medical writings. Handling the end to end data extraction, transformation, enrichment and storing the same for business use is the end goal of all Healthcare companies across goal.

South-East-Asia

Client

Healthcare Technology Solution Provider

 

CLIENT

Our client is a Healthcare Technology Services company from South-East Asia. They provide hospitals, clinics, labs, and healthcare units with the best records and data management services. They provide software modules and applications that can organize your medical notes, records, reports, and opinions with seamless integration, efficient monitoring, and enhanced insights.

We were tasked by the clients to find an intelligent end-to-end full fledged integrated solution to manage the medical unstructured document digital transformation pipeline. The goal also includes the problem of converting the paper-based (scanned, printed and handwritten) medical documents into an organized and properly formatted digital archival form.

 

CHALLENGE

The challenges faced by us are:

  • Problems with format identification - The formats used for internal and external documentation in the medical field can differ excessively, since the medical field has many sub-fields each with its own structural difference. Text, images, scanned documents, statistical inputs, etc are some of the type of data sources medical industry deals with.

  • Difficulties with the terminology, formulae and the stenographic quality of writing - Medical documentations are filled with complicated languages and terminologies where even a simple mistake could have grave repercussions. Generalised Language Intelligence Artificial Intelligence solutions fails to address the problem statements in medical industry, like spell-check, language vocabulary,  language understanding, entity understanding and processing.

  • Complex handwritten content standards are the by far the most commonly know issue - The style of writing in handwritten documents are optimized for speed, so it could be a little difficult to parse the script used for medical literature and documents which are nearly stenographic in nature.

  • Security and Confidentiality of Documents and Records - Most medical documents are protected by a legal clause dependent on their country of origin, it is extremely important to make the extraction clear but also prevent it from ending up in the wrong hands.

 

SOLUTION

Custome Med IDP solutions built from scratch are as follow -

In-house OCR

To address the problem related with image documents containing text information, our solution for the maintenance, archival, and security of medical records and documents was to devise an in-house Optical Character Recognition system. An Optical Character Recognition system extracts data and formatting from documents and files with extreme precision which then could be compiled digitally. OCR technologies exist as individual products in the market but their dependency on generalization and cloud-based retrieval would have a deep negative impact for our wants and needs. An in-house OCR, one that is developed specifically engineered to circumvent the aforementioned challenges presented by us is the perfect fit for our use case. They are tailor-made yet not restrictive since they can intelligently adapt to various styles of documents and could be updated with new paradigms as per requirement. With an adaption of state-of-art positional extraction, form extraction, table extraction and raw text extraction techniques we have developed in-house OCR to process scanned or printed documents. The OCR solution could be tweaked to store the data in any format necessary for the clients. This maintains cross-compatibility across all platforms and formats. Since, the in-house OCR is typically akin to a closed-source application; our data is protected from external tampering.

Med language editing  

We also built a language editing solution mainly for complex medical language understanding, language spell correction and recommendation (including procedures, medicine, unseen medical taxonomies), the solution is powered by state-of-art Transformer architecture, fine-tuned on custom data.

NLP modules

Several NLP enrichment  solutions are also developed and added to enrich the extracted medical information before storing in the knowledge-base.

Information transformation & storage are deployed and managed on a managed cloud services, and all data transfers happens in an in-house information PCDpt (PreCompressedDatapacket) packaging format for optimised data transfering .

Medical_record_digitization.png
med_OCR.png

In-house OCR flow

 

RESULT

  • We see a huge cost saving benefit for the company, as we have migrated their AWS services solution to in-house self controlled and managed solutions across IDP flow.

  • Using our in-house OCR model, we can definitely better the customer relationship focuses of our client.

  • Quick retrieval, secure storage, tremendous scalability - all point to an overall ease of engagement for our clients and theirs.

  • Doctors and patients alike benefits from this integration with their current enrichment platform, boosting sales and service productivity.

Ready to put AI to work for your business?

Make a plan and understand your ROI before you start implementing AI. 
Don’t fall into the trap most companies fall into. 
Take the first step—Get in touch today.