April 8, 2024
This is overdue, but… Day 2 of the PMSA Global Summit continued to offer insight into the pharmaceutical industry's challenges and opportunities that I still think about.
This was my first time attending PMSA Global, and also the conference's debut in India. Day 2’s sessions provided more insights, data points, and projections that set the tone for the industry’s trajectory.
From a global healthcare access perspective, Karthikeyan Chidambaram, Head of Global Data Assets Management at Roche, laid out a startling comparison: the average U.S. oncologist sees 24 patients per day, while their Indian counterpart sees 40 patients an hour!
This revelation led me to ask:
Inspired by retail and tech, Roche created flexible, cost-efficient Experience Centers. With the first one in Hyderabad, each center will be repurposed as training and meeting centers with the latest AR/VR technology to help scale medical education.
On the topic of AI, the conversation this day centered on experimentation. Unlike typical pharma analytics summits and conferences, where AI presentations are more marketing than learning, the PMSA event started with the customer's needs and expectations. Another discussion centered around bias and the dangers of overfitting data, with a show-your-work attitude.
The key takeaway is that RAG (retrieval augmented generation) still hallucinates, no matter which LLM you use. The last highlighted carefully engineered prompts, demonstrating how a state-table can pull geography, language, and user preference into the prompt frame using langchain.
As a community, we are smart enough to know these systems will transform the industry. Still, no industry-led generative solutions presented are HCPs or patients-facing… yet, and instead focused on tools for the internal team.
Panel Discussion: Global Capability Centers: The Way Forward
Moderator: Manish Mittal, Managing Principal, Axtria; Panelists: Neha Agarwal, Global Head Field Operations, Insights and Commercial, Novartis; Ankit Bose, Head of AI, NASSCOM; Karthikeyan Chidambaram, Global Head - Data Assets Management, Roche Pharma; Omkar Guru, Vice President, Healthcare & Life Sciences Analytics, Genpact; Mamata Kulkarni, Global Head - Insights & Analytics, Novartis; Mohit Sood, Regional Managing Principal, ZS
Given the conference in Delhi, the closing panel reviewed India’s role powering Global Capabilities Centers (GCCs). The critical 20% turnover rate, framed talent management as a river that must be continually navigated.
The key takeaway is to embed documentation and training into applications for seamless knowledge transfer. This mirrors our own workflows at CareSet, too, and where it doesn’t, it will. Similarly, Novartis shared they have an intensive training program meant to take a non-pharma top talent from zero to contributing, in as little as three weeks.
Lastly, the keynote from ZS CEO Pratap Khedkar painted a daunting picture of the current state and future of healthcare technology. He set the state with a statistic that medical knowledge doubles every 73 days. Within this overwhelming tide of information, the ZS CEO sees opportunity for AI's evolving role in diagnosing and patient care management. He cited a Nature article that reported that Google’s AI exhibited superiority in diagnosis and bedside manner, and provided a beacon for the future application of AI in healthcare.
This, when coupled with AI's potential in streamlining drug development and reducing clinical trial failure rates, Pratap foresees a way to upend Eroom’s law, which states that the cost to commercialize a molecule doubles every 9.5 years. He also highlighted something we’re currently focused on at CareSet, which is the fact that 11% of clinical trail sites never enroll 1 patient. At CareSet, we will work with CROs to select better sites, and/or work with the Medical team to help identify referral sources to ensure sites are more efficient, saving time, money, and lives.
The imperatives:
1) Talent and Process: Keep a squad that is cross functional that can own innovation from invention to adoption. There is a long standing rift between analytics and IT, and this is the opportunity to remove the division and get the cross functional team to think about adoption, trust, and management.
2) Adoption: what am I designing for? Accuracy and quality? NO! Design for user experience. A 10% error rate on a very usable system will always scale.
3) Trust: trust is about uncertainty. Any ambiguity is radioactive and will kill by installing an irresistible bias against you.
4) Mindset: Don't build “bridge” products because the river will always move.. build ferries and think about how to remain agile, while remembering anything you build will likely be a bridge. Buy infrastructure from partners (like CareSet :) that can keep up.
Thanks for reading, and let me know if you have any questions or comments!
