PARIS: An artificial intelligence (AI) program developed in China that analyzes test results, health records and even handwritten notes diagnosed of childhood diseases with the same precision as doctors, researchers said Monday (Feb. 12).
As reported in Nature Medicine: from flu and asthma to life-threatening pneumonia and meningitis, the system matched or consistently outperformed primary care pediatricians.
Dozens of studies in recent months have detailed how AI is revolutionizing the detection of diseases, including cancers, genetic disorders and Alzheimer's.
AI-based technology learns and improves in a way similar to that of humans, but has a virtually unlimited capacity for data processing and storage.
"I think he will be able to do most of the work that a doctor does," senior author Kang Zhang, a researcher at the University of California, San Diego, told AFP.
"But AI will never replace a doctor," he added, comparing the relationship with a self-contained car that remains under the supervision of a human driver.
"It will simply allow doctors to do a better job in less time and at lower costs."
The new technology, said Zhang, is the first in which AI absorbs unstructured data and the "natural language" to mimic the process by which a doctor realizes what is wrong with a patient.
"You can imitate a human pediatrician to interpret and integrate all kinds of medical data (patient complaints, medical history, blood and imaging tests) to make a diagnosis," he said.
The system can be easily transferred to other languages and configurations, he added.
By comparing hundreds of bits of information about a single patient with a large amount of acquired knowledge, the technology discovers links that previous statistical methods, and sometimes doctors of flesh and blood, overlook.
JUST IN TIME
To train the proof-of-concept system, Zhang and a team of 70 scientists injected more than 100 million data points from 1.3 million visits of pediatric patients at a major reference center in Guangzhou, China.
The AI program diagnosed respiratory infections and sinusitis, a common sinus infection, with an accuracy of 95%.
More surprising, Kang said, he did also with less common diseases: acute asthma (97%), bacterial meningitis and chicken pox (93%) and mononucleosis (90%).
Such technologies may be arriving just in time.
"The variety of diseases, diagnostic tests and treatment options have increased exponentially in recent years, which makes the decision-making process for doctors more complicated," Nature said in a press release. .
Experts who did not participate in the research said the study is further evidence of the expansion of AI's role in medicine.
"The work has the potential to improve medical care by helping the clinician make quick and accurate diagnoses," said Duc Pham, professor of engineering at the University of Birmingham.
"The results show that, on average, the system performed better than younger doctors."
"But it will not replace the clinicians," he added.
Machine learning, which forms the general rules based on specific examples of training, "can not guarantee 100% correct results, no matter how many training examples they use".
Tools based on artificial intelligence for diagnosis abound, especially to interpret images generated by machines, such as MRI and CAT scans.
A method presented last month in the United States to detect lesions that can lead to cervical cancer found precancerous cells with 91% accuracy, compared to 69% for physical exams performed by physicians and 71% for tests of conventional laboratories.
According to a study in the Annals of Oncology, an application for cell phones based on AI technology overtook dermatologists experienced in the distinction of potentially cancerous skin lesions from benign lesions.