Although research into machine use in medical studies has grown significantly in recent years, improvements in the clinical use of such data remain limited, according to a study published in npj Digital Medicine.
Machine learning (ML) is a promising but controversial tool for healthcare providers. The study raises growing interest about the possible use of ML in clinical settings, but also notes that appropriate guidelines should be used to ensure effective use. Recent research has shown that bias within artificial intelligence (AI) algorithms can cause health disparities.
The authors of the present study found that at each step of the research process, potential challenges and biases may be introduced that limit the clinical use of ML in clinical practice. Problems may arise from the start, depending on how the data for this study is collected, how the data sets are created and distributed, and what biases may be present in the databases themselves.
When data is tested, other challenges arise from selective targeting selection, preventing inappropriate testing procedures, selecting appropriate metrics, and adopting some of the best mathematical methods, the researchers point out.
In the publishing phase, certain incentives may affect the use of the information presented. For example, authors may use high-level language and mathematics to impress other scholars, leading to a lack of clarity and omission of important details. The pressure to publish paper with “novel” techniques and good results can also lead researchers to use more sophisticated methods. All of these factors reduce the frequency of a given study, which is the key to determining whether the results are consistent and effective for further use.
To address these challenges, researchers suggested raising awareness of data limitations, promoting the use of advanced systems to test machine learning, and improving publishing practices around reporting and transparency.
In addition to these concerns and concerns about the use of AI in general medicine, there has been significant success in ML recently.
One new study suggests that machine learning models can help diagnose complications of abdominal hernia surgery with high accuracy. Overall, these models predict a hernia recurrence at 85 percent accuracy, occurring at the surgical site with 72 percent accuracy, and a 30-day hospital stay at 84 percent accuracy. As well as improving patient outcomes, research shows that a 1 percent reduction in hernia recurrence rate, these examples can help you, could save the US $ 30 million health plan.
Another new machine learning algorithm has helped doctors alert patients to high-risk colorectal cancer patients. The algorithm uses such factors as age, sex, and total blood counts for an outpatient patient to determine which patients are at high risk for colorectal cancer. Nurses can use an algorithm to schedule colonoscopies for these patients. Of the 68 percent of patients who underwent colonoscopy during the study, 70 percent had significant findings during the procedure.
