Oncology vertical goes to PMSA

Deepika Sinha and Jeevan Elanchelvan, leaders in the oncology vertical at ZS, attended the PMSA Conference held in Delhi from March 6th to 8th, 2024. The Pharmaceutical Management Sciences Association (PMSA) conference is a non-profit and volunteer-based organization led by pharmaceutical analytics practitioners. Typically held in the US and Europe, this year it took place in India for the first time. They were particularly excited about this opportunity because the conference focuses on the latest advancements in analytics, and being in India for the first time made it a once-in-a-lifetime opportunity for them.

[Jeevan]: Hi Deepika, could you explain to our readers why the PMSA conference is important for oncology practitioners and why they should consider attending it?

[Deepika]: As expected, the summit focused on Generative AI (Gen AI) and scalable AI solutions. Oncology has its own nuances, such as a constantly evolving landscape, existing data gaps, and commercialization challenges. As an oncology practitioner, I was particularly interested in solutions that could have a significant business impact for our clients.

[Jeevan]: I agree. While many good ideas were discussed, not all of them were implemented in the context of oncology. However, there is no reason for oncology pharmaceutical companies to not implement them. We had multiple keynote sessions led by our clients and vendors, including our CEO, Pratap, who did an excellent job discussing the impact of AI. What were your key takeaways from the keynote sessions?

[Deepika]: All the keynote sessions were interesting, each offering unique takeaways. Here are my key insights::

1. Healthcare practitioners face significant challenges, including receiving nearly 80 alerts from EMRs leading to information overload, medical data doubling every 10 years leading to data assimilation complications, and 43% of US oncologists spending 20+ hours per week on documentation and administrative tasks. AI can operationalize these tasks, allowing HCPs to focus on more important aspects of patient care
2. 80% of clinical trials fail due to operational and patient recruitment challenges. Gen AI can identify potential recruitment issues at trial sites by predicting patient drop-off and deviations in the patient journey.
3. This one is my personal favorite, where one of the speakers presented an intuitive way of looking at AI solutions:
   - Patient-based AI solutions focus on early detection or screening, improving accuracy, reducing diagnostic efforts, and ensuring patients remain on therapy by predicting drop-off.
   - Decision support-based AI solutions assist physicians in accurately diagnosing patient cancers, selecting the best regimen, and generating structured clinical notes in real-time.
   - Generative AI can mine insights and significantly improve productivity by working on unstructured data.

[Jeevan]: There is a lot of excitement around Generative AI, and we saw multiple presentations during the conference on Gen AI. What use cases do you see as valuable for oncology clients?

[Deepika]: I see Gen AI having numerous applications across the healthcare value chain from providers to patients to pharma. While Gen AI is slated to increase productivity and reduce cost, it can also be used for deeper insights and better decision-making. Some examples of use cases include:

1. Providers
   - AI tools such as Gen AI chatbots and mobile applications can be leveraged to track patients’ treatment adherence and manage toxicities related to oral chemotherapy agents. This information can be relayed in real time to the HCP for timely intervention.
   - Gen AI integrates qualitative and quantitative data to understand each HCP’s motivations and preferences, influencing downstream promotional efforts.
2. Patients:
   - Gen AI analyzes patient call data to uncover unmet needs across stages of treatment journey.
   - Gen AI can predict patient drop-off and study unstructured data to enhance classical AI-based models, helping to understand the reasons behind the drop-offs.
3. Drug Discovery & Clinical Development:
   - Gen AI tools can document protocols, create trial reports, and more, reducing medical writing time by 30%. Examples include drafting design concept sheets, statistical analysis plans, and clinical study reports.
   - Gen AI-based tools can read patient clinical trial data to predict drop-off and create a digital twin for synthetic control arms and pre-clinical trials. This use case can significantly reduce the costs associated with clinical trials.
   - Gen AI can generate synthetic data such as protein-protein interactions to validate targets, facilitate disease heterogeneity understanding, and predict gene or lead prognosis and toxicity by predicting ADMET properties.
4. Marketing:
   - Gen AI can tailor digital marketing campaigns by creating variants of existing content and then personalize them based on HCP behaviors. It can simulate HCP/patient reactions to content and recommend enhancements.
   - Gen AI can analyze calls from reps to identify areas of improvement and coaching.
5. Medical
   - Gen AI assistance to call center and MedInfo agents can help deliver more consistency and a quicker time to issue resolution.
   - Gen AI can summarize publications to assist in clinical inquiries from the HCPs and create first draft emails, significantly reducing effort.
6. Supply Chain and Manufacturing
   - Gen AI alerts local supply shortages and demand spikes, while recommending efficient distribution plans and inventory allocations.

[Jeevan]: While there was excitement with Gen AI, there were also many solutions pertaining to Classical AI. Can you please talk about how Oncology clients should be thinking about it?

[Deepika]:

1. One key theme that emerged was deploying AI at scale, with multiple clients building capabilities to solve AI-based business questions at a much more rapid pace than before. The first step is to understand the business problems that various functions want to solve and to build a framework to create solutions for these problems. The solution includes readily available data, features that can be quickly built for different business problems, deployable models that can be operationalized (including ML operations), and a reporting capability to disseminate the outputs.
2. At this juncture, I also want to highlight ZS’s solution: Marketing n = 1. They started the presentation by showing a tombstone for traditional segmentation. Initially, I found it hard to believe, but by the end of the presentation, I was convinced it was the future. We do omnichannel next best action at the individual HCP level but the focus of this solution is more upstream by leveraging AI to identify the beliefs, drivers, and barriers of HCPs to create individual message plans.

[Jeevan]: Thanks, Deepika, for sharing your insights on the conference. To conclude, where do you think the future of AI in oncology lies?

[Deepika]: AI is here to stay, whether it’s classical AI or Generative AI. We know from multiple sources that HCPs will increase the utilization of clinical decision support services, in-silico models will help accelerate clinical trials, and in a more futuristic scenario, generative AI can identify hypotheses for testing and automate the initiation of clinical trials. Our clients can identify areas where AI can create business impact and enable scalable AI-based solutions, from drug discovery to commercialization. At ZS, we can help clients in this journey.

Read more insights from ZS.