What patients and the industry are asking for: healthcare, personalized
Patients have long been asking for a future of healthcare where digital healthcare technology rapidly evolves to enable medical teams to:
- Diagnose and treat medical problems faster and more seamlessly
- Provide a comprehensive approach to treating patients across systems and reports
- Alert patients with preventive recommendations to lower the risk of adverse health outcomes in the future, like heart disease or diabetes
- Provide a connected end-to-end experience to meet them where they are with their health needs, from initial encounter to discharge
In order to align generative AI’s capabilities with patient expectations and business needs, healthcare business leaders looking at this from a revenue cycle perspective will need to develop strategic plans for applying generative AI in both front-end (scheduling, patient registration, eligibility and authorization) and back-end (claims management, reimbursement) processes. Generative AI can even review its own outcomes, using data from patient satisfaction surveys, net promoter research, and data analytics to ensure that organizations are on the right track with the initiatives implemented.
Testing a use case for ChatGPT in healthcare
Publicis Sapient recently used ChatGPT to develop a series of prompts based on an individual’s health situation geared toward finding out if the tool could support better health outcomes and found the following trends to report on:
Dialogue flow is limited, and ChatGPT has a tendency to take action straight away
- Responses are extremely verbose; ChatGPT produced 40,000 words in response to 97 prompts
- Responses tended to be generic and lacked nuance; however, it responded with empathetic phrasing and apologized when it got things wrong
More data equals stronger analysis
In the case of one interviewee, the team set out to test whether ChatGPT could help improve their health outcomes, by crafting a series of prompts focused on:
- How the individual could lead a healthier lifestyle
- How good ChatGPT is at recommending preventive health activities
- What advice it had for the employee to support a lifestyle condition, such as GERD (acid reflux)
The subject in this example was a white, middle-aged male living in the city with a sedentary lifestyle, a stressful desk job, two teenagers, aging parents and several health conditions.
While the input from ChatGPT was initially generic, the more data that was provided (for example from the Apple Health app) the more ChatGPT provided stronger answers (e.g., when answering a question about a potential colonoscopy). This speaks to the need for effective implementation of patient data in generative systems to create more robust opportunities for use as the technology continues to develop.
Solving for the current limitations of generative AI
AI and machine learning are currently being used in healthcare; however, there are a few hurdles to overcome before generative AI will become fully embedded in future healthcare ecosystems.