Yitu Healthcare claims over 90% accuracy when diagnosing a range of pediatric diseases using a natural language processing system
China’s Yitu Healthcare has achieved accuracy rates of as high as 97% in a research project to diagnose pediatric diseases using natural language processing (NLP), a branch of artificial intelligence (AI) to help computers make sense of written and spoken language.
This makes Yitu’s AI capabilities on par with that of doctors when reading electronic health records and generating patient diagnoses. The firm’s research, which was conducted with the Guangzhou Women and Children’s Medical Centre and other research institutions, was recently published in Nature Medicine, a peer-reviewed medical journal.
Using 101.6 million data points from over 1.36 million outpatient visits over an 18-month period, the researchers built a NLP-based diagnostic system supported by a data model that integrates clinically relevant information and prior medical knowledge.
The primary diagnoses included 55 diagnostic codes for common diseases in pediatrics, covering a wide range of pathologies. The diagnostic system achieved robust performance common conditions as well as potentially life-threatening ones.
Besides recording a 97% accuracy rate in diagnosing acute asthma exacerbations, the researchers claimed accurate diagnoses of bacterial meningitis (93%), varicella (93%), influenza (94%) and roseola (93%).
Ni Hao, president of Yitu Healthcare, the healthcare arm of Yitu Technology, said the breakthrough has proven that AI technology can help doctors deal with large amounts of data, as well as diagnose uncertain and complex medical cases. “Pediatric diseases can be tricky for doctors. An AI assistant will profoundly improve the diagnosis process and increase efficiency,” he added.
In a clinical setting, the system can be used during triage to diagnose medical conditions based on a patient’s medical history and vital signs, as well as physical examinations. Yitu said besides helping doctors to avoid misjudgment and biases in their diagnoses, the system will also ease the shortage of experienced pediatricians in China.
The use of AI in medicine has been growing in the past few years. In 2017, Australia’s Icon Group started tapping IBM’s Watson for Oncology to identify cancer treatment options and drugs, supported by research papers and medical information from medical journals and textbooks.
The tool further ranks evidence-based treatment options across seven types of cancer, linking to peer-reviewed studies and clinical guidelines. With machine-learning capabilities, it continuously learns over time, based on previous interactions with its users.
Berg, an AI startup, is also getting in on the act, with a drug discovery platform that combines patient data with clinical and demographic information to provide the doctors with recommendations on treatment plans.
You can copy the link to share to others: http://www.4g2m.com/node/583