A profession under threat? AI system outperforms junior doctors in diagnosing childhood illnesses

The medical profession is a highly revered one, and rightly so. It is the profession that has significant impact on life and death, ability to bring back to life, as it were, in a physical form.

However, a new study is bringing to the fore issues that were hitherto not envisaged, and thus challenging the enigma that surrounds this profession.

A new Artificial Intelligence (AI) system has been found to perform better than junior doctors in spotting childhood illnesses, researchers have found.

The programme, developed by researchers from China, was able to diagnose with approximate degree of accuracy of senior paediatricians in primary care setting.

Different range of diseases including asthma, life threatening pneumonia, and even meningitis were diagnosed accurately by the AI system.

There have been series of evidence that demonstrated that AI system – Machine Learning Classifiers (MLC) – is able to make diagnoses using clinical images.

However, the new AI system is the first of its kind in that the computer tool is able to filter through unstructured data , including case notes, to arrive at accurate diagnosis.

The developers used an automated machine learning system to sieve through vast amount of electronic health records (EHR) to extract clinically pertinent information to make decisions on diagnosing illnesses.

The system is able to make scientific deductions, similar to that of the “hypothetico-deductive” reasoning of physicians in reaching conclusions about associations that previously were not detected by statistical methods.

In reporting their findings, the researchers noted: “In total, 101.6 million data points from 1,362,559 pediatric patient visits presenting to a major referral center were analyzed to train and validate the framework. Our model demonstrates high diagnostic accuracy across multiple organ systems and is comparable to experienced pediatricians in diagnosing common childhood diseases.” 

It is remarkable that the system can mimic human pediatricians to use patient data – complaints, their medical history and examinations, blood and imaging investigations – to make diagnosis.

The system will never replace a doctor but could help ease the work of a doctor, especially in processing large amount of data and could prove most useful in areas in dire shortage of human resources for health.

“Our study provides a proof of concept for implementing an AI-based system as a means to aid physicians in tackling large amounts of data, augmenting diagnostic evaluations, and to provide clinical decision support in cases of diagnostic uncertainty or complexity. Although this impact may be most evident in areas where healthcare providers are in relative shortage, the benefits of such an AI system are likely to be universal, ” the researchers concluded.

Please read more about this study in Nature Medicine.

Samuel Abiona

Samuel Abiona

Samuel Abiona is a medical doctor by training and a writer by passion. Samuel holds a postgraduate degree in Public Health. He believes that communicating medical knowledge goes beyond writing technical reviews. Samuel thus uses his expertise in public health and health systems research to transmit technical information for both academic and general audience. Please email samuel.abiona@essaysinhealth.com to contact this author directly or use the contact page and your information will be passed on to him.