No matter the specialization or the healthcare facility, patients rely on doctors to review their symptoms, order the correct diagnostic tests and reach the best solution possible in the shortest amount of time. With many conditions, a delayed diagnosis or a missed diagnosis can result in a worsening affliction, lengthy hospital stay or a challenging prognosis. Many physicians now rely on computer programs to aid the speed and accuracy of their diagnoses.
Unfortunately, these advances in medical technology might not represent the immediate step in the right direction that many people had hoped for.
Several years ago, the American Medical Association published a study that, for many, was a sobering dose of reality. The call for better artificial intelligence (AI), stronger options and a better decision-tree was based on findings in the study – namely, that human doctors outperformed their computer counterparts by a wide margin. During the study, the human doctors reached a correct diagnosis 72% of the time while the computer diagnostician reached the correct diagnosis only 34% of the time.
Even though many considered this a win for the doctors, it is worth noting that the doctor reached a misdiagnosis more than a quarter (28%) of the time. These misdiagnoses can represent deadly infections, worsening conditions and unnecessary pain for the patient.
What can be done?
Like any study centering on self-driving vehicles, the healthcare profession’s technology must advance in iterations. Diagnostic AI could eventually reach a level of accuracy rivaling or surpassing human doctors, but medical professionals should still only use it as a diagnostic aid. Computer findings could potentially assist a doctor in reaching a diagnosis or shedding light on uncommon possibilities.
In recent years, computer diagnostics has moved beyond a simple decision tree into numerous pages filled with several fields of information. From measurements such as height, weight, blood pressure and temperature to the associated family history, algorithms increase in depth of detail and intensity. For now, however, doctors must use these tools in conjunction with an accurate history and educated hypothesis to avoid a deadly misdiagnosis.