Artificial intelligence is already transforming everything from filmmaking to cybersecurity, and it may also be on the verge of producing huge medical advances that have baffled scientists for decades.
According to an article in Fortune, the application of artificial intelligence in medicine has increased in recent years, particularly in disease diagnosis. A growing number of doctors rely on deep learning, a machine learning technique modeled on artificial neural networks that learn by example, like human brains, to help detect potentially life-threatening conditions that are easy to overlook, such as cancer, heart disease, and even asymptomatic cases of COVID-19.
Alzheimer’s is a debilitating disease that causes permanent cognitive loss and dementia. Treatment and reliable early detection have evaded medical doctors in the century since the disease was discovered.
Researchers at Massachusetts General Hospital recently tested deep learning techniques in Alzheimer’s detection and discovered that not only was deep learning more accurate than comparative A.I. models that weren’t trained to analyze multiple variables simultaneously, but it was also able to identify Alzheimer’s cases regardless of factors that typically complicate early-onset detection, such as a patient’s age. The findings were published in a study last week in the scientific and medical journal PLOS ONE.
The researchers developed a deep learning model using tens of thousands of brain scan pictures from over 10,000 individuals with and without Alzheimer’s disease. The model was then compared against real-world clinical data on Alzheimer’s diagnoses. The deep learning model identified Alzheimer’s cases with a 90.2% accuracy rate, or five percentage points higher than the simpler A.I. models that did not utilize deep learning. The AI model performed better regardless of when, where, or at what age patients were initially diagnosed with Alzheimer’s.
Matthew Leming, a research fellow at Massachusetts General Hospital and the study’s principal author, said in a statement: “This is one of the only studies that attempted to diagnose dementia using routinely obtained brain MRIs.” The cross-site, cross-time, and cross-population generalizability of our results supports the practical application of this diagnostic method.
77%, according to a 2017 study, are the human clinical detection rates for Alzheimer’s disease. A 90% accuracy rate in Alzheimer’s diagnosis would be light years ahead of these rates. While OpenAI, Microsoft, and Google’s A.I.-powered search engines have dominated recent artificial intelligence news due to their potential to revolutionize search and the way we work, machine learning could have potentially vital uses in medicine.
According to a study published in December by the U.S. Department of Health and Human Services, more than 7 million emergency room patients are wrongly diagnosed annually. This study indicated that about three million emergency room patients are burdened with harmful effects due to a misdiagnosis, while over three hundred and seventy thousand suffer a permanent disability or death.
According to the non-profit Organization tforBetter Diagnosis in Medicine, reducing inaccurate tests and treatments as well as malpractice lawsuits resulting from misdiagnoses might result in annual savings of approximately $100 billion.
A number of the same difficulties with A.I. that have been discovered elsewhere, such as the possibility for factual errors and racial prejudices, have also surfaced in medical studies, according to doctors and clinicians. A literature analysis of A.I. in medical diagnosis published last year indicated that the technology has potential in domains such as cancer, diabetes, and Alzheimer’s diagnosis; nevertheless, additional study is necessary to increase A.I.’s accuracy in recognizing medical disorders.
Alzheimer’s is one of the most difficult diseases to forecast and detect, but if future research makes AI and deep learning more commonly employed in diagnosis, it might be a game-changer. Alzheimer’s disease is the most prevalent form of dementia among the elderly, affecting around 44 million individuals globally. Yet, Alzheimer’s disease is simply one manifestation of a vast family of dementia-related illnesses that are frequently misdiagnosed.
2017 research of over 900 individuals indicated that up to one-quarter of Alzheimer’s patients were incorrectly diagnosed, with false positives and false negatives occurring in about equal proportions. Alzheimer’s propensity for misdiagnosis is largely attributable to the fact that many of its symptoms match with those of other common neurological illnesses, such as Lewy body dementia and frontotemporal dementia.
According to the American Academy of Neurology, the likelihood of misdiagnosis increases with age, and Alzheimer’s disease and other dementing conditions “may be easily misinterpreted in the elderly.”
Forecasting whether or not a patient will develop Alzheimer’s is just as difficult as identifying the disease, as over 90% of Alzheimer’s cases are deemed “sporadic” – occurring in persons without a family history of the disease. Due to these challenges, there are essentially no good early screening models for Alzheimer’s, and the majority of patients are detected after the onset of brain damage symptoms.
The Massachusetts General Hospital study did not address whether deep learning may aid in Alzheimer’s disease prediction, but other studies seem to indicate that artificial intelligence could play a significant role there as well.
The University of Florida reported last week that an artificial intelligence model was able to utilize electronic health information to identify patients with a high risk of acquiring Alzheimer’s up to five years before a diagnosis. A.I. models could aid in early disease identification and lower the disease’s long-term severity, the researchers observed. Nevertheless, further testing is necessary before doctors begin using AI prediction tools.
