Remodeling AI-powered Key Opinion Leader (KOL) Search Function






Efficiency in Search Function
Medscrape Case Study Header Image

A proposed answer to navigate scores of data to find relevant KOLs

Healthcare providers must scrouge through multitudes of data to discover and identify the relevant Key Opinion Leaders (KOL)s within their respective therapeutic areas. A Seattle-based tech startup was looking to build an MVP to combat this time-consuming task of manually scouring and aggregating information. This novel way of retrieving and presenting data would create a wave of transformation in systematically viewing healthcare data.


The client proposed integrating information from multiple sources (viz. publications, clinical trials, new APIs, etc.) to deliver all in a single platform. They also wanted the data to be represented to end users through robust visualizations, i.e., knowledge graphs, for ease of comprehension and comparison. They approached us with the problem, requesting to build modules for their MVP’s new feature – UX/UI, front-end, and back-end.


Rapid Application Development, Product Engineering, Data Engineering


Python, Javascript, Flask, React, GraphQL, MySQL, Neo4j, AWS Elastic Bean Stalk


Seattle, USA

Barriers to a practical resolution of the issue

In addition to working in collaboration with the client in a different time zone, the engineers had to learn new coding languages and techniques to create the latest tool in a record one-month timeframe.


  • The project involved quickly and efficiently incorporating many JSON files into Neo4j, a graphing database. These files included healthcare practitioners, organizations, open payments, and taxonomy data. 


  • Analytical graphs were to be created for the dashboard through real-time aggregation queries on the millions of Neo4j data sets.


  • A ‘Global Search AutoComplete’ feature was to be implemented. This would allow the user to input any character, and if any property of any node contained those keywords, the search would quickly return that data.


  • Various knowledge graphs with different custom designs were part of the required outcomes.

The innovative solutions put forth

Despite challenging time constraints on the project, the team successfully delivered results within the given timeframe. They closely collaborated with the client while creating the new software. They made new approaches and learned new skills to complete the project.


  • They first converted the JSON files into CSV format. They had to create separate CSV files for nodes and relationship data. Then, they implemented a Neo4j admin tool to import the new CSV files. It allowed them to insert around 70 million nodes and 200 million pieces of relationship data in five to six minutes instead of hours.
  • Since the queries for generating the graphs were mostly OLAP queries, they denormalized the data and precalculated expensive aggregations. They then stored the data and aggregations on new nodes and relationships.
  • The team implemented a full-text search on the nodes using Lucene indices and a distance-similarity calculation, and they could fetch the results in little time.
  • To streamline the knowledge graph generation, the team used only one component each for tables, line charts, and bar charts. The team used a general approach by ignoring the designs attached to each graph type and dynamically passing those to the desired components.

Modules implemented

Global Search Engine

With global search engine, Customers can search healthcare providers by single input text which can be an address, product, manufacturer, or medical school related to providers. The global search is performed with millions of data in real-time. This also facilitates auto-completion of input text.

Healthcare News

News related to Healthcare providers such as products, life science firms, and specialties from various sources can be browsed.

Twitter Module

Users can see the recent tweets of a particular healthcare provider. They can also see the trending hashtags and popular twitter handles per therapeutic area.

High Level Design Architecture

Need a custom software application for your business?

We at TechVariable do acknowledge that one size will not fit all. Hence, we work in collaboration with you to identify, analyze & then develop a solution that fulfills your needs. Either we will define the functional scope of your project to estimate the timeline and budget or you can create your own agile team from among our resources.
estimate project

The results acquired through undertaking the search function

  • The new Global Search Autocomplete feature allowed users to search millions of KOL data, publications, clinical trials, and related industry information in a single search box. These advanced aggregations occurred in real time, constructing relevant and actionable user knowledge graphs.
  • The team also created a module linking healthcare practitioners’ Twitter accounts to search results. This enabled end users to see relevant KOL tweets and hashtags currently trending. Additionally, the results showed similar Twitter accounts for the appropriate therapeutic field.

The new search tool was less expensive and used more unique modules than its competitors, giving them an edge in the industry.

Previous slide
Next slide

See how our solutions are making a difference in healthcare