AI-powered digital algorithms for discovering potential KOLs have become an important topic in pharma. From pre-launch R&D stakeholders to later medical insurance access, hospital access, and key figures in volume-based procurement, using big data analysis tools and scientific algorithms for intelligent recommendation is truly powerful.
So what is a real big data algorithm?
Let us start with a real case:
A pharma company's human albumin project team once entrusted us with a broad-market target KOL discovery project.
If this were an ordinary drug with clear target indications and concentrated departments, it would not have been difficult. The key challenge was the special nature of human albumin for critical care fluid replacement. This meant there was no target indication; any disease that could lead to critical illness might involve its use. There was no specialized academic society, and no guiding guideline consensus. In what seemed like a situation with no clear starting point, the goal was to find precise experts across many departments.
For MeDomino's experienced project team, however, the difficulty was manageable. After the project quickly started, the team worked with the client's medical affairs department to study authoritative materials and identify KOL characteristics as fully as possible. They eventually developed a feasible solution and achieved efficient, precise targeting.
With big data, the keys to success in this project were three words: comprehensive, precise, and fast.
01 Comprehensive
A large and comprehensive physician database is the foundation of the whole project. It is like searching for the world's most beautiful pearl in a small pond versus the ocean. The better option is obvious.
We once compared MeDomino's physician database with statistics from the health commission yearbook and found that our database covered 98% of physicians in secondary and tertiary hospitals, while coverage of physicians in primary and unrated institutions exceeded 90%. Maintained over six years, this physician database is clearly industry-leading. Such comprehensive and clean data has become the guarantee behind MeDomino's high-quality project delivery.
02 Precise
Precision has two meanings here:
First, physician data matching must be highly accurate.
Matching logic is one of the most difficult steps in HCP profiling because it involves hospital and department aliases, physician institution changes, and matching diseases mentioned in data with expert interest. It requires technical extraction and long-term manual maintenance. During projects, we continuously optimize matching logic. To date, rough statistics show more than 200 segmented algorithms, with both programmatic recall and accuracy exceeding 95%.
Second, the target KOLs must be accurately located.
This requires a full evaluation system, especially a quantitative model for evaluating cross-department and cross-specialty KOLs' influence and focus in a certain disease or treatment method.
It also requires considering comparability across different data types, such as whether HCP characteristics and various data from different departments and specialties are comparable. Only when data types and departments or specialties are comparable can the model support refined screening in the project.
03 Fast
For this type of project, the most time-consuming part should not be data integration. It should be early solution planning.
Spend more thought on identifying target diseases and designing step-by-step solutions. With a complete physician database, data extraction becomes a one-step process. It is not only precise, but also more efficient, saving a great deal of time. This also represents a major advantage of big data algorithms, which old-fashioned manual search methods cannot achieve.
After screening through the three key points above, we believe companies can move beyond traditional methods and use big data algorithms to identify cross-department experts in broad-market settings, efficiently helping enterprises expand coverage. The same methodology also applies to broad disease fields, certain medical devices, and other KOL identification scenarios. Their shared feature is that KOL departments are not concentrated, which also shows the breadth of the market. What we need to do is identify more experts through data as early as possible, so good products can benefit more patients.
However, as one company said, fake big data companies are common. They claim to have data and AI algorithms, but in reality they do not even have an independently operated physician database. How can that be convincing? This pitfall is actually easy to avoid. The strongest test is whether the provider supports verification and spot checks. We are confident on this point. If you do not believe it, feel free to contact us, and we will show you a demo and let you verify it.
