Search
Close this search box.
Search
Close this search box.

Case study

Deep neural network based system generating medical documentation abstracts and voicebot-performed medical interviews abstracts supporting medical diagnostics during medical consultancy.

Challenge

Deep neural network based system generating medical documentation abstracts and voicebot-performed medical interviews abstracts supporting medical diagnostics during medical consultancy.

Share this:

Organisation profile

telDoc operates on the medical market as:

Non-public health care facility providing medical services in a stationary and remote manner using chatbots and / or voicebots
an entity that develops modern IT solutions dedicated to healthcare.

The most important projects implemented by telDoc:

  • Development, verification and implementation of a triage service based on artificial intelligence – integrated with medical teleconsultations and voice assistants and a virtual contact center (R&D project);
  • Complete SONAR system for mass testing of asymptomatic people for the presence of the Sars-Cov-2 virus,

    implemented and delivered at the request of the nencki institute of experimental biology (Instytut Biologii Doświadczalnej im. M. Nencki PAN); as part of the project, applications were developed to carry out mass tests for the presence of the SARS-CoV-2 virus using the SONAR method, i.e. (1) “covid” voicebots, (2) Mobile applications for scanners for collection points and laboratories, (3) SONAR portal, in which the user can check the test result and his visit history, integrating voicebots and mobile applications.

    telDoc conducts talks with many commercial entities (diagnostic laboratories, insurers, medical facilities or technological entities) for the implementation of medical voicebots.

Solution

The goal of the project is to develop a system for automatic generation of comprehensive medical synopses (in Polish) based on analysis of:

  • medical interview with the patient conducted by the voice bot
  • digital medical documentation available at the patients hospital/ medical facility

Tasks, methods of achieving goals, main functionalities of the project’s outcome -development of machine learning models generating the abstracts of the medical documentation containing critical, from the point of view of the medical diagnostics, information. The quality of the generated abstracts will be measured by calculating the ROUGE values – parameters measuring the similarity between the generated and the reference abstracts prepared by the qualified medical personnel.

  • ROUGE-1: 38, ROUGE-2: 23, ROUGE-L: 37 (abstractive and extractive abstracts, generated from the provided, digital medical document)
  • ROUGE-1: 40, ROUGE-2: 22, ROUGE-L: 36 (abstractive and extractive abstracts, generated from the provided, digital 5/1.1.1/2020 POIR.01.01.01-00-1204/20 medical document)

No other solution available on the market at present time achieves above mentioned qualities because of:

  • The Polish market – no tools available for synopses generation in Polish
  • The foreign market – all tools available are in English, their implementation is not possible due to the language differences.

The system will be dedicated in the first stage to 3 specializations:

  • General surgery
  • Neurosurgery
  • Orthopedics and motor organ traumatology

The primary focus groups are:

  • Hospitals (public and private), outpatient entities (direct recipients)
  • Doctors (indirect recipients – the main users of the new system), hospital/ outpatient clinic patients (indirect recipients)

Results

Project in progress.

Get in touch

MIM Solutions can help develop AI in your company, please contact us and we’ll talk you through it.