The Signal: Healthcare AI Gets More Practical
The Signal is MKG’s look at AI developments shaping healthcare and life sciences, with a focus on what they could mean for biopharma and healthcare organizations.
This week, AI showed up in more of healthcare’s day-to-day work, including clinical search, prescribing, prior authorization, physician education, and research. As these tools become more common, questions about accuracy, oversight, and accountability matter more.
Pharma + Healthcare
Clinical Search Gets a Cited Answer
Elsevier will begin rolling out ClinicalKey’s Ask AI feature to U.S. customers in September, reaching more than 1,500 hospital sites. The tool draws from ClinicalKey’s journals, guidelines, books, and clinical overviews, provides citations, and does not use the open web. It is another sign that HCPs may increasingly use AI to find and explore clinical evidence rather than searching for individual documents.
Source: Elsevier
AI Can Now Help Prepare A Prescription
First Databank has launched Script Agent with Tebra, an AI tool that uses the clinician-patient conversation to prepare a prescription for the clinician to review, change if needed, and send. AI tools used during patient visits have largely focused on creating clinical notes. This extends that technology into prescribing while keeping the clinician in control. As AI becomes more involved in prescribing, the accuracy of the medication information behind these tools becomes even more important.
Source: First Databank
AI Takes on Prior Authorization
DrFirst launched an AI-assisted electronic prior authorization tool that uses information from the prescribing record to prefill payer questions for clinician or staff review. In a company-reported one-month pilot covering 150,000 pharmacy authorizations, it completed 63% of question sets without edits and recorded a median 49 seconds from open to submit. The early results are promising. The bigger questions are whether it can reduce administrative burden, lower denials, and help patients get to therapy faster.
Source: DrFirst
AMA Adds AI Ethics Training for Physicians
The American Medical Association introduced a four-module Ethical AI Use in Medicine CME series, developed with Duke Institute for Health Innovation and the Health AI Partnership. The case-based training covers evaluating and monitoring AI, clinical decision-making, mortality prediction, and ambient scribes. As physicians become more educated about how to evaluate AI, expectations around transparency, evidence, and the limits of AI-enabled tools are likely to rise.
Source: American Medical Association
AI Study Looks at Treatment Response in Pancreatic Cancer
A Caris-led study looked at 2,202 molecularly profiled pancreatic cancer cases and used AI to identify groups of patients whose outcomes differed depending on their first-line treatment. Because the study looked back at existing patient data and was missing some clinical information, the findings still need to be tested in prospective studies. The study shows one way AI could help identify treatment-response patterns and inform future biomarker research and clinical trials.
Source: npj Precision Oncology
Outside the Industry
ChatGPT Ads Expand Across Europe
ChatGPT Ads became available in 31 European markets on August 24, OpenAI’s largest expansion of the program so far. Ads appear only on Free and Go plans, remain labeled and separate from answers, and now include expanded targeting, conversion optimization, and measurement. As ads begin appearing within AI platforms, privacy, promotional rules, and a clear separation between paid content and AI-generated answers will be especially important for pharma.
Source: OpenAI
Google Tests Ads That Start Conversations
Google’s August Demand Gen update is testing ways for YouTube viewers to start conversations with brands in messaging apps directly from an ad. Google also expanded its AI video creation tools in Asset Studio. This could make it easier for someone to move from seeing an ad to having a direct conversation with a brand. It also raises practical questions around consent, review, and keeping approved claims consistent across formats.
Source: Google Ads
The Takeaway
AI is becoming part of more healthcare workflows, from clinical search and prescribing to prior authorization and research. As adoption grows, the bigger questions are whether these tools are accurate, transparent, and used with the right level of human oversight.