In recent years, artificial intelligence (AI) research has made great progress, and its application and positive impact in the medical field are becoming more pronounced. To date, perhaps the most advanced AI-based solutions are in diagnostic areas such as the analysis of radiological images. Another area of interest with high commercial impact is projects related to the identification of potentially therapeutic molecules. AI is also increasingly being used to improve quality of life and, more generally, quality of care and patient experience.
For the latter, tools like ChatGPT, which reached millions of users in just a few weeks, could play a role. Some early experience has documented using ChatGPT to support bureaucratic activities such as issuing medical certificates and reimbursement claims with health insurance companies. Other uses are for use in more complex activities such as triage. This means that access to services in resource-constrained settings may establish priorities based on a number of pre-defined criteria, or provide information about ongoing treatments and treatments. . At the same time, it collects direct patient feedback that can influence treatment strategies.
However, tools such as ChatGPT, and AI solutions in general designed as consumer products rather than for medical purposes, raise a series of ethical and legal issues. For example, in one of the ongoing research initiatives at the SDA Bocconi School of Management’s Government, Health and Non-Profit Department and his CEGAS Research Center in the Healthcare sector, I recently evaluated the possibility of inserting ChatGPT. I noticed that there is With the ability to create in the medical app, you can create a text designed to provide all the necessary information to a woman who has been diagnosed with breast cancer and is preparing for a course of treatment.
In a very short time, the chatbot generated plausible text. We are reviewing the essential contributions of clinicians and medical professionals for publication. The question is whether the information thus generated is accurate and fair enough for medical application. Sensitive contexts such as health protection.
Other challenges related to the use of ChatGPT in the medical field include potential privacy violations and, unfortunately, potential risks due to the transmission of sensitive data that may be disclosed to third parties in the future. will be There is also, to date, a lack of informed consent forms suitable for various uses of AI.
CERGAS is currently working on another solution designed for women who are scheduled for mastectomy followed by breast reconstruction. In her four-year project in Europe, Cinderella aims to improve the satisfaction, and thus the mental health and quality of life, of patients undergoing breast cancer surgery by automating the aesthetic assessment of surgical outcomes. is. Predict outcomes prior to intervention to promote women’s active and conscious participation in treatment selection.
With so many possible surgical alternatives, today it is difficult for women to decide which type of surgery can give the best aesthetic results in their case. In order to automate and make these assessments objective, an algorithm (BCCT.core) created by INESC TEC in Porto and the Breast Department of the Champalimaud Foundation in Lisbon has been used for several years to extract chest photographs of patients undergoing surgery for breast cancer. began to be classified according to the following criteria: to aesthetic results organized mainly on the basis of geometrical indicators. For example, the software can autonomously establish breast volume and bra cup size from a photo. Needless to say, the images used were anonymized with the consent of the woman whose chest was taken.
In order to limit the unfairness of the answers provided by the software, we extended the image database used to train Cinderella’s AI to cover a range of initially underestimated subjects in terms of skin tone, transparency and body type. A woman had to be included. Thanks to recent advances in so-called deep learning that allow optimization of image recognition, this advanced software will be able to provide personalized predictions for individual patients about the outcome of various surgical approaches. .
But there must be one element at the root of everything. It is the effective and fair implementation of AI solutions centered around people’s needs and appropriate safety and ethical standards. This is a fundamental condition for unlocking these innovation potentials by improving the quality of AI solutions. Medical assistance and community welfare.
Courtesy of Bocconi University
