PostgreSQL Goes Multi-Model: Graph, Vector, and SQL
A second case study of the clinical trials
A clinical trial is a type of research that evaluates the safety, efficacy, and effectiveness of experimental drugs, medical devices, or procedures in humans. They are crucial for advancing medical knowledge and improving patient outcomes. However, they can be extremely costly. A 2020 study by Moore et al. revealed that the median cost of 101 drug trials was a substantial US$48 million. And one of the major cost factors is the recruitment of enough qualified volunteers. Strict eligibility criteria narrow the potential participant pool, while limited awareness of clinical trials among eligible patients further hampers recruitment efforts.
In my last article Clinical Trial Search with Google Spanner: Graph, SQL, Vector, and LLM All in One Query, I outlined the development of an information system that enables patients to search for valuable data on drugs, diseases, mechanisms of action (MOA), and clinical trials. The system was built on the fully managed, multi-model Google Spanner database. Freed from the constraints of single-model databases, I was able to store data flexibly in a combination of graph, table, and vector formats. This approach simplified the data modeling process, reducing the time, effort, and complexity involved. Furthermore, users could…
