Bio
I am a clinical scientist by training, and my time in hospital-based mental health settings shaped my career early on. I saw firsthand how disconnected data and tools keep patients and providers from seeing the full picture. The same silos can cloud decision-making in clinical development alike. I started my AI journey in large pharma building clinical trial endpoint monitoring solutions so patient exacerbations could be tracked in near real time, then moved to work with Real World Data, unifying claims, EHR, and clinical notes at scale. At Norstella, I work directly with customers to close gaps in their data landscape so customers can traverse a unified data ecosystem with ease and focus on developing the next best treatment for patients.
What makes Norstella’s AI different?
Two things: the data and domain expertise. Norstella’s data ecosystem spans clinical intelligence, regulatory, news and insights, and commercial, as well as various types of RWD. This gives you a 360-view of patients, providers, investigators, and the market – all harmonized and connected from day one. I haven’t seen this combination anywhere else. Just as important is the domain expertise: our analysts and scientists bring disease and clinical knowledge to make sure data is fit-for-purpose. This allows us to build AI solutions that are fit-for-purpose, powered by best-in-class data.
What can pharma teams do now that was not possible two years ago?
When I worked on a top 20 pharma data science team, we still sometimes wrestled with data and infrastructure silos. That’s changing – unified data solutions like Norstella’s empower pharma teams to connect their clinical and operational data to third-party data at the enterprise level. This means you can build AI/ML solutions with more precision and at a larger scale within your organization.
How can AI accelerate the journey from pipeline to patient?
AI is a broad term covering many machine learning techniques, not just LLMs. For drug discovery, you may want graph neural networks to model molecular structure and predict drug-target interactions. For market access, natural language processing applied to unstructured payer and policy data can help surface insights faster. To move quickly, we must stay nimble, educated, and sophisticated in our understanding of AI – that’s one of my guiding principles. Where I’d call out large language models specifically is democratization of data access: with the right data foundation, experts can now ask questions in their own words, for example, the way they would in a clinical trial protocol, and AI can reliably translate those to derive insights from complex data. That excites me as it saves hours and days from data wrangling and is ultimately a less error prone process as it comes to mapping clinical concepts to relevant data elements and underlying coding schemas.