An artificial intelligence (AI) model with high accuracy, comparable to that of experienced pediatricians in diagnosing common childhood diseases, is reported in a paper published online this week in Nature Medicine. These findings provide proof-of-concept for implementation of an AI-based system to help physicians tackle large amounts of data, augment diagnostic evaluations, and to provide a clinical decision support in cases of diagnostic uncertainty.
Medical information has become increasingly complex over time. The range of diseases, diagnostic testing and treatment options has increased exponentially in recent years. Subsequently, clinical decision-making has also become more complicated.
Kang Zhang and colleagues developed an AI-based model that applies an automated natural language-processing system that uses deep learning techniques to identify clinically relevant information from electronic health records. This model can search electronic health records and unearth associations that previous statistic methods have not found. In total, 101.6 million data points from 1,362,559 pediatric patient visits to a major referral center in Guangzhou, China were analyzed to train and validate the framework.
The model demonstrated a high level of accuracy for diagnoses compared with initial diagnoses by an examining physician. It also performed well when diagnosing two important categories of disease: common conditions, such as influenza and hand-foot-mouth disease, and dangerous or life-threatening conditions, such as acute asthma attack and meningitis.
The authors conclude that this type of AI framework may be useful for streamlining patient care, such as by triaging patients and differentiating between those patients who are likely to have a common cold from those who need urgent intervention for a more serious condition.
Epidemiology: A website to assess COVID-19 event risk in the US in real timeNature Human Behaviour
Materials: Making strong bio-based replacements for plasticsNature Communications
Pterosaur teeth reveal dietary preferencesNature Communications