The goal of this project is to develop an automatic claims categorization module for openIMIS based on state-of-the-art Artificial Intelligence (AI) algorithms, standards, and methodologies which will drastically reduce the manpower, resources and time required to review a reimbursement claim. As a contribution towards achieving Universal Health Coverage, the Swiss Tropical and Public Health Institute (Swiss TPH) designed and developed the health insurance software ‘openIMIS’. Other partners joined later such as the IT firm SolDevelo for specific work areas, thus bringing together state of the art Public Health and Software development expertise.
This project aims to develop an Artificial Intelligence (AI) based claims categorization prototype application to automatically update the claim’s status to accepted, rejected or to be further analyzed by a medical expert. The process will be divided into three phases:
-
Research phase: Swiss TPH will work closely with GIZ Nepal to undertake the research and development of the AI algorithm for claims categorization based on anonymized openIMIS claim data and associated entities (insuree, health facility, diagnostics, medical items, and services). It will be crucial to have access to a database with already categorized digitized claims (by a Medical Officer) and to identify the key input variables to be processed by the AI algorithms. The overall objective of the AI component is to achieve claims categorization performance similar to the human expert results. Several classification methods (supervised and unsupervised) will be considered to fit the specific AI model to be elaborated. To make the solution available for different contexts and insurance entities, HL7 FHIR will be used for the input and output data format of the AI model.
-
Development phase: SolDevelo will develop an openIMIS module that will integrate the AI algorithm and will create the requisite links and transitions for the module activation. Moreover, an extension of the Claim module will be developed to allow the Medical Officer to check the quality of the AI categorization results and adjust, if necessary, thus improving more the AI categorization model.
-
Testing phase: Swiss TPH and GIZ Nepal will organize a workshop in Nepal and run the AI module in real case scenarios to validate the research and development phases.
Building on solid implementation and software maintenance expertise over the past 7 years, we are currently extending the partnership to Nepal Health Insurance Board/GIZ Nepal, who is an existing openIMIS implementer for the Informal Sector that can already make categorized claims data available. With this background, we are confident that the automatic claims categorization module can be built and validated in the proposed time frame. The Digital Square investment would allow us to realize the automated (and AI-enhanced) claims categorization that will add substantial benefit to the health insurance system community.
Swiss Tropical and Public Health Institute (prime organization)
Swiss TPH is a leading institute in global health with a particular focus on low- and middle-income countries with a staff strength of over 850 staff from 80 different nations, currently active in 300 projects across 100 countries.
Swiss TPH will provide technical project management, expertise on health financing, openIMIS, and AI algorithm development, drafting of business and technical specifications, supporting the system and architecture design.
Swiss TPH’s relevant experience includes involvement in the design and implementation of the Insurance Management Information System (which is the genesis of the openIMIS Initiative) since its inception in Tanzania and has supported its implementation in a number of countries. Swiss TPH is currently implementing two projects at scale for the deployment of insurance schemes through openIMIS, in Tanzania and Cameroon, and two openIMIS pilots in Chad and Democratic Republic of Congo. In addition, Swiss TPH is actively involved in the development of openIMIS as part of the Implementers and Developers Committees of the openIMIS Initiative. Further, Swiss TPH is also implementing projects with AI-based on supervised machine learning to improve Clinical Decision Support Systems.
Qualifications of key members of the proposed project team:
-
Dragos Dobre:
-
IT System Architect (PhD. in Automatics and Computer Science, MSc. in System Engineering) at Swiss TPH
-
Management of software life cycle (from specification to development and deployment)
-
Design, development, and maintenance of open-source applications
-
openIMIS design, development and implementation experience since 2018 in implementation sites of openIMIS - Tanzania, DRC, and Chad
-
Project Coordinator experience across international teams for GIZ openIMIS mandate
-
OMG-Certified Systems Modeling Professional™
-
-
Siddharth Srivastava:
-
Health Financing Specialist (MSc. in Operational Research) at Swiss TPH
-
Over 10 years of experience in health insurance projects in Lower Middle-Income Countries (LMICs: India, Nepal, Bangladesh, Cambodia, Cameroon, Tanzania, Kenya, Malawi; backstopping roles in Chad and DRC)
-
Insurance Information systems design (capturing and documenting user requirements) and implementation (including capacity building) experience
-
IMIS/openIMIS design and implementation experience since 2013 in all implementation sites of openIMIS - Tanzania, Cameroon, Nepal, DRC and Chad
-
Project Manager/Project Coordinator experience across national teams for GIZ openIMIS mandates and insurance projects in Nepal, Cameroon, and Kenya
-
-
Simona Dobre:
-
Data scientist (PhD. in Automatic Control and Signal Processing, specialized in modeling, identification and data analysis)
-
Over 10 years of experience in data analysis in interdisciplinary domains
-
Project Manager/Project Coordinator experience across international teams
-
Machine Learning and Deep Learning certification
-
SolDevelo
SolDevelo is a dynamic IT company founded in 2009 with over 80 employees and focuses on delivering high-quality software and innovative solutions.
SolDevelo will be responsible for the software integration of the AI module with openIMIS.
SolDevelo is currently involved in several openIMIS projects, including HL7 FHIR module development, openIMIS integration with OpenMRS project, maintenance and support project, and enhancing the security of the Microsoft solution. SolDevelo has been involved in many opportunities that required skill sets relevant to this particular project, especially through opportunities like OpenMRS (core contributors), HL7 FHIR implementation (OpenMRS Sync 2.0 module), nationwide micro-service based implementations (OpenLMIS), nationwide OpenHIE architecture based implementations (National Health Infrastructure project with such components like OpenELIS, DHIS2, OpenMRS and many other HIE compatible applications, health standards-based workflows for the Client Registry, Facility Registry, Health Management Information System, Shared Health Record, and Interoperability Layer).
Qualifications of key members of the proposed project team:
-
Kamil Madej
-
Senior-level Java Developer/Team Leader (BSc. Engineering) at SolDevelo
-
Working in international teams for various projects/clients, like:
-
openIMIS
-
OpenMRS
-
MOTECH
-
Terre des hommes
-
Connect for Life
-
-
Performing code review
-
Creating high-level designs using tools for wireframing
-
Leading several frontend and backend development teams
-
Nepal Health Insurance Board / GIZ Nepal
Nepal Health Insurance Board is the implementer of openIMIS in Nepal from 2016.
Together with GIZ Nepal, they will be responsible for the appropriation of categorized claim data, the validation of the anonymized data set for AI analytics and the testing of the openIMIS AI module developed under this project.
Tanzania, Nepal, Cameroon, Democratic Republic of Congo, Chad
Comments
Alignment of claims with Product Master Data
This is a strong concept note! At VillageReach we are working on an OpenHIE Product Registry solution called PCMT, Product Catalog Management Tool. We submitted a concept note (https://applications.digitalsquare.io/content/openimis-product-registry-interoperability) and we would be interested in offering collaboration or assistance with interoperability as part of your project, particularly in the area of assuring claims are aligned with the product master data. Please email PCMT@villagereach.org if you are interested in exploring this.
Scalability
In your final proposal you should clearly state how you want to make the solution scalable e.g. how will a local implementer (e.g data scientist) be able to chose (choice of algorithm), configure (choice of attributes) and train (with local historic data) the ai component to their needs. Also state if you want to implement features for contious quality monitoring of the ai results.