Pneumonia, a condition where the air sacs in the lungs become swollen and infected, is the cause of approximately 14% of fatalities in children under five years old, resulting in the loss of more than 700,000 young lives annually, according to the World Health Organization (WHO).
Although most adults can recover from pneumonia, especially with medical intervention, it can be extremely dangerous for children and individuals with weak immune systems - especially in underdeveloped regions where access to treatment is limited.
Undoubtedly, an early, prompt, and cost-effective diagnosis of pneumonia would significantly impact the outcome. Currently, diagnosis of pneumonia typically involves blood tests and chest scans, which can be both costly and time-consuming, and a physician must suspect pneumonia to request them.
Detecting Pneumonia Thru Coughs
However, there may be a recognizable symptom of pneumonia that could be used to diagnose the disease without additional testing - coughing. Pneumonia coughs are distinct from other coughs because the lung inflammation associated with the disease modifies the sound produced when coughing. It may be difficult for humans to discern this difference, but with the aid of specialized electronic equipment and a machine-learning algorithm, it can be accomplished.
At a meeting of the Acoustical Society of America, Jin Yong Jeon from Hanyang University presented a machine learning algorithm that uses sound analysis of a cough to diagnose pneumonia. The algorithm is trained to differentiate between coughs caused by pneumonia and those caused by other factors, and once it has developed enough proficiency, it can be used for diagnosis.
This is not a novel concept, as several other research teams have been exploring the use of technology to diagnose pneumonia or monitor lung health, sometimes using a smartphone microphone. Researchers aim to create new, cost-effective methods of diagnosing diseases by combining low-cost sensors with sophisticated algorithms. Jeon and his team enhanced the recordings by incorporating room impulse responses, which assess the effect of a room's acoustics on different sound frequencies.
Promising AI with High Accuracy Rate
By incorporating these responses with the recorded cough sounds, the algorithm can perform more accurately in any setting. The research resulted in a 97.5% accuracy rate for the dataset. According to Jeon, automatic diagnosis of a health condition through analysis of coughing sounds captured daily will enable non-in-person treatment and reduce medical costs.
A company has already expressed interest in using the algorithm for remote patient monitoring. If the algorithm can work with simple equipment in any setting, it could significantly impact the millions of patients who contract pneumonia each year. Jeon stated that their research team is working on automating manual processes to enhance convenience and practicality, as reported by ZME Science.
The use of AI for diagnosing pneumonia is just one example of its growing role in disease diagnosis. Algorithms are proving valuable in helping doctors diagnose various conditions, from heart disease and Parkinson's to childhood diseases. The goal of this technology is not to replace medical professionals but rather to assist them and make it easier for them to reach a diagnosis. This technology has the potential to make a significant impact on early diagnosis and treatment across a variety of diseases. However, it is important to note that the recent study presented at a conference has not yet undergone peer review.
RELATED ARTICLE : Explainable AI Accurately Labels X-Ray Images of Five Chest Pathologies Equivalent to Seven Human Experts
Check out more news and information on Artificial Intelligence and Pneumonia in Science Times.
