Researchers aimed to evaluate the efficacy of the OpenAI technology and offer guidance into its future implementation within health care services.
ChatGPT was found to be effective at assisting with patients’ medication therapy management (MTM) in 39 out of 39 cases, according to research published in the Journal of the American Pharmacies Association.1 However, it was less successful in its assessments as specific cases became more complex.
MTM services include reporting health status assessments, creating treatment plans, modifying therapy, and providing resources, among other benefits to patients. For the study, researchers identified specific functions of ChatGPT and how it could be used to assist in MTM services.
One function is assisting health care providers in drug-drug interactions.
“A drug-drug interaction occurs between 2 or more drugs, a drug and food, or supplements that a patient takes, or a drug and the health conditions that a patient may have,” wrote the authors. “These interactions have the potential to cause significant harm to patients and should be monitored carefully.”
To yield the most successful results, researchers tested ChatGPT’s effectiveness in 3 criteria: ability to identify drug interactions, successfulness in recommending alternative services/medications, and appropriateness of creating management plans.
Researchers gathered 39 specific MTM cases and tested ChatGPT’s effectiveness with the help of 2 clinical pharmacists to confirm its results. They then ranked each case by complexity—simple, complex, and very complex—to further assess ChatGPT’s successfulness in assisting each specific case. After recording ChatGPT’s responses, the pharmacists compared its assessment with real answers provided to past MTM cases.
ChatGPT was 100% successful in assessing all 39 MTM cases | image credit: ParinPIX / stock.adobe.com
Researchers included 13 cases for each complexity and assessed ChatGPT’s responses in each case, yielding valid results in all 39.
“A response was considered valid if ChatGPT could accurately provide 70% of the clinical points from each criterion when compared to the answer,” wrote the authors.
The artificial intelligence (AI) was able to determine interactions between drugs, diseases, substances, and supplements in all cases. However, despite reaching the required 70% in all 39 cases, researchers pointed out ChatGPT’s declining effectiveness as cases became more complex.
“ChatGPT can help health care providers avoid trial-and-error medication selection and reduce the risk of adverse drug events,” wrote the authors. “However, to obtain a correct response, it may require follow-up questions and specific clinical training and knowledge, unlike traditional MTM where a health care professional can perform physical exams and order lab tests.”
The successfulness of ChatGPT in this study can help medical practitioners understand AI’s specific uses for improving health care systems, especially in electronic MTM plans.
“With the help of AI, conducting MTM has the potential to become quicker and more efficient through improving medication safety, predicting health risks and outcomes across different populations, reducing time and expenses, monitoring medication adherence, and scheduling follow-up health care services,” continued the authors. “AI integration into the health care system continues to evolve the future of health care and provides new opportunities for future health care providers.”
But ChatGPT and other AI technologies do not come without their limitations. Two specific concerns with the integration of AI in health care systems are patient skepticism and issues in data security and privacy.
Regarding skepticism, the authors highlighted how the population of MTM patients could be wary about AI’s much-recent inception and implementation into general services, only entering the health care field in the 1970s.
Although still in its early stages of implementation, full AI integration in the health care industry is an increasing possibility.
“AI experts can help develop and refine algorithms, identify areas for improvement, and ensure the technology meets the needs of both providers and patients. The integration of AI and clinical pharmacy can lead to better patient outcomes, increased medication adherence, reduced health care costs, and improved patient engagement in their own health care,” concluded the authors.1
READ MORE: AI Can Improve Business Operations in Community Pharmacies and Benefit Patients
