Over the past decade, significant strides have been made in diagnosing and treating pulmonary hypertension, driven by updated criteria and new medications. However, timely diagnosis remains a major challenge, with many patients facing long delays, often waiting years and consulting multiple doctors before receiving an accurate diagnosis.
Journalist and healthcare writer Jared Kaltwasser has recently published an article on “Managed Healthcare Executive” titled “How Deep Learning and AI Could Fuel the Next Phase of PAH Detection” which cites recent promising research on this issue.
The most recent research Kaltwasser cites, titled “A Deep-Learning-Enabled Electrocardiogram and Chest X-Ray for Detecting Pulmonary Arterial Hypertension” was conducted by a group of researchers in Taipei, Taiwan, and was published in the “Journal of Imaging Informatics in Medicine”, on August 13, 2024. The researchers developed a deep-learning model (DLM) that combines electrocardiograms and chest X-ray data to detect elevated pulmonary arterial pressure. The model was tested with data from two hospitals, collected over the course of 11 years, parsing electrocardiograms and chest x-ray records from more than 30,000 patients each. The model demonstrated strong performance, with high sensitivity and specificity in detecting elevated pulmonary arterial pressure. It also had a 98% negative predictive value, meaning it was very reliable in ruling out the condition in those who didn’t have it.
Corresponding author Wen-Hui Fang, M.D. and colleagues said use of their model could be an important screening tool. They suggested that if implemented in clinical practice, the model could help identify which patients might need more invasive testing, such as right heart catheterization. However, Fang and colleagues also emphasized the need for further studies to explore the correlation and interpretability of the relationships identified by these models.
A previous 2022 study titled “A framework of deep learning networks provides expert-level accuracy for the detection and prognostication of pulmonary arterial hypertension”, was published in the “European Heart Journal – Cardiovascular Imagining” in November 2022. The researchers used data from 450 pulmonary arterial hypertension patients, 308 with right ventricular dilation without pulmonary arterial hypertension, and 67 healthy controls to train deep learning models linking echocardiographic images with estimated right ventricular systolic pressure. Their algorithm achieved 97.6% accuracy and 100% sensitivity in detecting PAH. While promising, the researchers emphasized that the model is not yet ready to replace human experts but could serve as a useful screening tool.
Citation 1st study
Liu, PY., Hsing, SC., Tsai, DJ. et al. A Deep-Learning-Enabled Electrocardiogram and Chest X-Ray for Detecting Pulmonary Arterial Hypertension. J Digit Imaging. Inform. med. (2024). https://doi.org/10.1007/s10278-024-01225-4
Citation 2nd study
Gerhard-Paul Diller, Maria Luisa Benesch Vidal, Aleksander Kempny, Kana Kubota, Wei Li, Konstantinos Dimopoulos, Alexandra Arvanitaki, Astrid E Lammers, Stephen J Wort, Helmut Baumgartner, Stefan Orwat, Michael A Gatzoulis, A framework of deep learning networks provides expert-level accuracy for the detection and prognostication of pulmonary arterial hypertension, European Heart Journal – Cardiovascular Imaging, Volume 23, Issue 11, November 2022, Pages 1447–1456, https://doi.org/10.1093/ehjci/jeac147
Photo credit Growtika/Unsplash
