With great promise to transform the rare disease diagnosis scene, artificial intelligence (AI) has emerged. Using huge databases from genomics, imaging, and electronic health records, artificial intelligence systems can identify patterns that even experienced doctors often overlook (1).For those with orphaned diseases, who sometimes have to bear years of uncertainty, this capacity to shorten the diagnostic journey is quite crucial. But while AI has advantages, its abrupt involvement in diagnostic procedures presents major ethical and clinical problems necessitating more investigation. Most artificial intelligence algorithms' interpretability poses a major difficulty. Usually black boxes, deep learning algorithms provide diagnostic results without a clear justification. This hurts openness and may compromise the mutual decision-making process between doctor and patient (2). In medicine, especially when rare diseases are being diagnosed, where patient involvement and customized treatment are vital, trust in medical decision-making grounds decision-making. Bias in the training data for artificial intelligence is another critical problem.Learned from data from high-income nations, the majority of algorithms do not take under resourced populations into consideration (3). This could compromise diagnostic accuracy in ethnically and geographically varied populations and so worsen existing health inequities. For instance, in underrepresented communities (4) using recent artificial intelligence models applied to the diagnosis of genetic disease, performance gaps were observed. Furthermore worrisome is the possibility that doctors will increasingly rely too much on artificial intelligence systems. A 2024 multicenter simulation study showed how clinicians can become too dependent on AI output, even when it goes against clinical intuition (5).Particularly in rare disease environments when subtlety and world understanding are essential, this might cause loss of critical thinking ability and aid in unfavorable patient outcomes. Though AI has the potential to revolutionize the identification of rare diseases, its application in clinical practice must be consistent with the tenets of openness, equity, and ethical responsibility. Independent real-world practice testing, inclusive training datasets, and ongoing clinician education are required to make sure that technical developments result in actual patient benefit.
Abdullah et al. (Tue,) studied this question.