Stock photo of a doctor speaking with a patient.Jacob Wackerhausen/STOCK PHOTO/Getty Images
An artificial intelligence (AI) system is assisting some individuals in obtaining diagnoses after enduring years of unexplained ailments, according to a recent investigation.
Scholars from OpenAI and Boston Children’s Hospital re-examined the existing genetic information of 18 pediatric patients—many now adults—utilizing a newly developed AI model, thereby resolving cases that had baffled medical professionals for years.
The research group anticipates that their model will aid numerous American children who are affected. The investigators noted that one in every 10 individuals in the U.S.—exceeding 30 million people, with half being children—suffers from a rare disease.
The researchers are of the opinion that their models could potentially help in diagnosing the most uncommon diseases and also accelerate the diagnostic process.
The findings were disseminated on Thursday in the New England Journal of Medicine.
Stock photo of a doctor speaking with a patient.Jacob Wackerhausen/STOCK PHOTO/Getty Images
For Kyra, who is 28 years old and was one of the participants in the study, the extremely rare diagnosis of Myofibrillar Myopathy (MFM) was reached after almost two decades of ambiguity. MFM encompasses a category of genetic conditions leading to progressive muscle weakening.
“It seemed incredibly unreal at the moment because I simply did not anticipate receiving an answer within my lifetime, and I believe my family did not anticipate it either,” Kyra shared with ABC News.
Even though the condition is currently without a cure, “we finally obtained that clarity and sense of resolution,” she stated. “At the very least, it’s gratifying to have a name for it.”
The research does not suggest that AI can supplant physicians or genetic specialists. While the model proposed potential answers, medical experts rendered the definitive diagnoses, and each diagnosis was verified by an accredited clinical laboratory prior to informing the families.
That distinction is significant because AI systems are susceptible to errors and misinterpretations of data, as indicated by the researchers.
Within this study, the AI functioned more like an additional set of eyes for specialists, aiding them in analyzing extensive volumes of intricate data in approximately six to 10 minutes per case, according to the researchers.
Catherine Brownstein, a principal investigator in the study and a research associate in the genetics and genomics division at Boston Children’s Hospital, informed ABC News that by leveraging AI, “we can allocate our human expertise to more specialized tasks, such as scrutinizing the data, rather than pursuing improbable diagnostic avenues.”
Kyra, the patient, expressed her belief in the potential of AI for cases like hers, while also emphasizing the need for careful supervision.
“I believe that for the purpose of supporting researchers in their endeavors, particularly when they involve such intricate and complex situations as seen in this study, it can indeed be a highly valuable instrument,” she remarked. “However, concurrently, I think we must proceed with considerable caution.”
As per Brownstein, safeguarding privacy is fundamental to the application of this type of technology.
“We are not removing any human oversight mechanisms here,” Brownstein stated. “A human must review every action taken by the AI.”
The study also highlights the value of re-evaluating past genetic test results. Given the rapid advancements in science, a finding that seemed inconclusive years ago may become clearer as researchers identify new genes, enhance their methods for analyzing genomic data, and deepen their understanding of how genetic variations impact health.
“A genetic test that currently yields a negative result might not do so in the future,” Brownstein informed ABC News.
Stock photo of a doctor holding the hands of a young patient.Thianchai Sitthikongsak/STOCK PHOTO/Getty Images
Brownstein mentioned that the swift progress in genetic discoveries can make it challenging for medical practitioners to continuously re-examine old, unresolved cases, but AI might offer assistance.
“The human genome is being deciphered more and more each day,” she commented. “But AI excels remarkably in this area.”
In Kyra’s specific situation, only 60 patient instances of MFM have been documented in published literature, according to CureMFM13, a non-profit organization dedicated to addressing the challenges associated with the condition.
The research team acknowledged certain limitations of the study. Investigators retrospectively examined existing cases, thus the study cannot definitively confirm whether the tool would perform identically in real-time scenarios.
Furthermore, the quantity of newly identified diagnoses was modest, and the study did not quantify whether the AI tool resulted in time savings, cost reductions, or improvements in patient care.
The subsequent phase, as outlined by the authors, involves evaluating the methodology in more extensive prospective studies across multiple healthcare institutions.
For Kyra, the diagnosis did not erase the past years of uncertainty, but she indicated that it provided her and her family with elements they previously lacked: a definitive name, a sense of closure, and a connection to others living with the same rare condition.
She stated that while AI might help researchers discover answers more rapidly, the human aspect of healthcare remains paramount.
“When dealing with health matters that profoundly impact one’s life, one tends to desire the presence of a human connection,” Kyra expressed. “You want to feel that individuals care, that they are attentive to you, and that you are more than just a medical case.”
Joshua Anthony, MD, MBA, is a psychiatry resident at Creedmoor Psychiatric Center and a member of the ABC News Medical Unit.
Sourse: abcnews.go.com