Pfizer Launches ‘Health Answers’ AI Tool: A New Approach to Consumer Health Information
Pfizer has introduced “Health Answers by Pfizer,” a new generative AI product designed to provide consumers with trustworthy information about common health questions. The tool functions similarly to ChatGPT, generating responses from “verified sources,” complete with citations, after a user submits a query. This approach offers a potentially faster alternative to searching for health information on platforms like Google.
However, Pfizer acknowledges that the responses are not medical advice. This raises concerns about the risk of users misinterpreting the information. LLMs, such as the one used by Health Answers, can sometimes generate inaccurate information, as opposed to human-edited content.
Initial Impressions and Observations
After testing the tool, Health Answers provided a variety of answers to a question about knee pain and suggested a doctor’s visit. In another instance, the AI tool flagged an unverified response when it lacked sufficient content from trusted sources to generate a verified answer per the platform’s standards, though, it still provided an answer.
When prompted with a question involving fentanyl, the AI correctly identified the risks. Notably, when asked the same question, the AI generated by ChatGPT provided an incorrect answer and highlighted the potential for confusion that these language models can cause.
Monetization and Data Practices
Pfizer states Health Answers is independent of its pharmaceutical business, but the company operates “PfizerForAll,” a consumer portal that connects users with telehealth providers who can prescribe medications. Links to Health Answers are featured on PfizerForAll, and Health Answers provides links to healthcare services such as BetterHelp, Sesame, and Zocdoc.
The platform’s website notes that Pfizer may earn a fee when users click on links. Additionally, Pfizer prompts users to create a profile and share health information to provide a personalized experience. Users can opt-in to receive marketing communications from “Pfizer Affiliate Partners” based on their use of the site, potentially including their shared health data. Pfizer warns that “certain information, such as health information, may be considered sensitive.”
Technology and Safeguards
Pfizer credits Microsoft for its assistance in developing the tool, and Microsoft has close ties to OpenAI. According to Pfizer, Health Answers uses an LLM with proprietary enhancements to optimize responses. To mitigate risks, Pfizer has implemented safeguards, including oversight from an independent scientific and technology advisory board, filters to ensure accuracy, clinician audits, and disclaimers throughout the platform.
News Briefs
HIMSS Conference Hit by Federal Official Absences
The HIMSS conference in Las Vegas saw the absence of numerous federal officials who were scheduled to attend. Due to event travel regulations implemented by the Trump administration, at least 46 officials representing health agencies like the FDA, CMS, and CDC were initially registered. Despite this travel pause, the conference is still underway, and an important conference for health IT professionals.
ASTP/ONC Faces Uncertainty Amidst Potential Cuts
Layoffs at the Assistant Secretary for Technology Policy (ASTP) organization have raised questions about its future, with rumors circulating about further government downsizing. ASTP, which oversees electronic health records and artificial intelligence within the federal health department, is less than a year old. The office’s current role is in flux, and there is also a new acting head.
DOJ Investigates Semler Scientific
The Department of Justice is probing Semler Scientific for possible violations of the False Claims Act. The company’s primary product is QuantaFlo, used to test patients for peripheral artery disease. The DOJ probe disclosed the investigation in an SEC filing that indicated settlement discussions with the government. Company officials stated in the filing that they would vigorously defend against any charges.
How AI May Warp Healthcare Incentives
Ken Mandl, director of the computational health informatics program at Boston Children’s Hospital, discussed how commercial interests can influence the development and application of seemingly-objective medical measures. This concept, termed “biomarkup,” applies to AI within healthcare. AI-based revenue cycle tools may push clinicians toward care pathways that are financially more profitable.
Mandl noted, “Wherever there are financial flows, there are going to be parties that are interested in optimizing those financial flows.”
