Patient Discharge Summary

Mahavir Dialysis Centre- Patient Discharge Summary using GenAI

  1. Home
  2. >
  3. Mahavir Dialysis Centre Patient...

Executive Summary

Founded in 2009, Mahavir Dialysis Centre is a prominent non-profit organization and a leading dialysis center in India. With a network of over 17 centres nationwide, the organization has received recognition from esteemed bodies such as the World Bank, WHO, and the DFID of the U.K. Mahavir Dialysis Centre is deeply engaged in charitable activities, driven by a mission to support patients suffering from chronic kidney disease (CKD) who require regular dialysis at affordable costs. To streamline their current manual and time-consuming discharge summary process, Minfy and Mahavir Dialysis centre collaborated to address their challenges.

Customer Challenges

Mahavir Dialysis Centre, a prominent healthcare provider, manages extensive medical documentation for patients' inpatient department (IPD) journeys. A critical task for healthcare providers is the creation of discharge summaries during patient discharges. The current manual process for generating these summaries is highly time-consuming, taking approximately 20 to 30 minutes per report. This manual approach imposes administrative burdens on healthcare providers and hampers their ability to focus on delivering quality patient care. Moreover, the time-consuming nature of manual report creation can cause delays in discharging patients, potentially impacting patient flow and overall operational efficiency.

Minfy’s Solution

Minfy developed a solution to automatically generate summarized medical reports from electronic health record (EHR) data. The solution consists of three main components: data ingestion and pre-processing, Large Language Model (LLM) fine-tuning and deployment, and an automated report generation pipeline.

Data Extraction and Processing

Patient data from the EHR system is stored in an Amazon S3 bucket. Amazon Textract is used to retrieve relevant textual data from the medical documents stored in this S3 bucket. The textual data extracted by Amazon Textract is passed to an AWS Lambda function for further processing. This Lambda function performs necessary data cleansing, formatting, and anonymization tasks on the extracted data.

AWS Lambda function is triggered to invoke a pre-deployed Large Language Model (LLM) endpoint for generating patient discharge summary reports. This Lambda function receives the processed data after the processing step, sends it to the LLM endpoint, and stores the generated reports in an Amazon S3 bucket. An EC2 instance with a FastAPI application serves as the user interface. Users can search for patient IDs, which triggers a pipeline to retrieve the corresponding patient's contents and generate the discharge report.

The Open-Source Large Language Model is fine-tuned on medical domain data and deployed as a SageMaker endpoint. This model is responsible for generating accurate and concise discharge summary reports based on the processed patient data received from the Lambda function. We chose Mistral 7B Instruct, an open-source Large Language Model (LLM), for our medical report analysis and discharge summary generation solution. This decision was driven by the model's capability to be effectively fine-tuned on medical domain data, enabling accurate comprehension of medical terminology and contextual nuances. Furthermore, Mistral 7B Instruct offers better customizability, seamless deployment flexibility on AWS SageMaker, and potential cost-effectiveness, making it a superior choice over proprietary models for this critical healthcare solution.

Amazon Service: S3
Rational: Highly scalable and durable object storage for storing medical documents from the EHR system and processed patient data.

Amazon Service: Textract
Rational: Leverages machine learning to accurately extract text, handwriting, and data from virtually any document, making it ideal for retrieving relevant textual data from medical documents.

Amazon Service: SageMaker
Rational: Used for fine-tuning and deploying the Large Language Model on medical domain data, enabling accurate and tailored discharge report generation.

Amazon Service: VPC
Rational: Provides a logically isolated and secure network environment for the EC2 instance and other components.

Amazon Service: Lambda
Rational: Serverless compute service used for data processing tasks on the extracted textual data and invoking the SageMaker endpoint for report generation.

Amazon Service:EC2 instance with FastAPI
Rational: Provides a user interface for searching patient IDs and initiating the report generation process

Results and Benefits

Mahavir’s collaboration with Minfy, leveraging strategic AWS services, resulted in substantial benefits:

The solution automatically generates accurate discharge summaries in less than 2-3 minutes, reducing reporting time by over 90%. It seamlessly integrates multiple Amazon services like Textract, SageMaker, Lambda, and S3 for efficient data handling, secure delivery, and interoperability among centers. A user-friendly web portal enables healthcare providers to quickly search patient records and download branded, customized reports, reducing front desk time by 70%. By automating the creation of discharge summaries, the solution allows providers to better allocate their time and expertise towards enhancing patient care, resulting in a 20% increase in the number of consultations conducted per day.

This website stores cookie on your computer. These cookies are used to collect information about how you interact with our website and allow us to remember you. We use this information in order to improve and customize your browsing experience and for analytics and metrics about our visitors both on this website and other media. To find out more about the cookies we use, see our Privacy Policy. If you decline, your information won’t be tracked when you visit this website. A single cookie will be used in your browser to remember your preference not to be tracked.