Artificial intelligence‐assisted automated heart failure detection and classification from electronic health records – ESC Heart Failure-Wiley Online Library

Posted By: admin May 7, 2024

Singapore – 7 May 2024 –  New research led by the University of Dundee [1], United Kingdom, demonstrated the feasibility of AI to automatically identify and classify patients with heart failure  from archived echocardiographic images with  Us2.ai software.

Heart failure (HF) is a highly prevalent yet under-diagnosed condition with high mortality and morbidity [2]. Echocardiography is a foundational investigation to diagnose HF and differentiate the types of HF i.e. HF with reduced (HFrEF), mildly reduced (HFmrEF) and preserved (HFpEF) ejection fraction [3, 4].

Electronic health records (EHRs) are an increasingly high-quality data source that can be used for the creation of pragmatic cohort studies [5], disease surveillance, case selection for clinical trials (RCTs) [6], and quality improvement initiatives [7]. The quality and quantity of EHR data are expanding and increasingly include EHR-linked biobanks [8-9] and EHR-linked imaging data [10].

This study aimed to identify and classify patients with HF from routinely stored EHR data, linked to Scottish Health Research Register (SHARE) [11] bioresource and echocardiographic data collected from the Tayside and Fife region of Scotland using a deep learning-based approach.

AI-automated analyses of DICOM echocardiographic images using Us2.ai software was combined with analysis of biomarkers from routinely stored plasma samples using Roche Modular E170 (Roche Diagnostics, Mannheim, Germany). AI image analysis accurately quantified both systolic and diastolic left ventricular function, as well as structural characteristics of left- and right- atria and ventricles. The research team demonstrated the feasibility of identification and differentiation of HF particularly distinguishing between types of HF at large scale in a streamlined time and cost-efficient manner.

” Our approach has potential clinical implications, especially in the precision required in HFpEF clinical trials and the broader context of heart failure diagnosis and surveillance. The automation our DL algorithms not only makes the diagnosis process more efficient than traditional methods but also paves the way for identifying heart failure cohorts more pragmatically. When combined with biobank data, such as that from the SHARE project, our methods hold the promise of accelerating biomarker validation and fostering innovations in drug discovery for heart failure treatment” said Dr Chim Lang, from the University of Dundee’s School of Medicine.

These data are further supported by prior evidence that AI-automated echocardiographic image analyses with Us2.ai is interchangeable with human experts [12], produces measurements comparable to gold-standard invasive hemodynamic filling pressures [13], is generalizable in both real-world and research cohorts worldwide [14], and has potential for mobile screening applications [15].

About Us2.ai
Us2.ai uses machine learning to automate the fight against heart disease. The company’s software tools improve clinical decision-making and cardiovascular research for clinical trials using echocardiography, the safest and most common cardiac imaging modality. Us2.ai connects institutions and imaging labs around the world on a platform of ready-to-use automation tools for view classification, segmentation and reporting of findings according to International Guidelines and recommendations.

References:

  1. This study was funded by ROCHE DIAGNOSTICS
  2. Lam, C.S.P., et al.,Heart failure with preserved ejection fraction: from mechanisms to therapies. Eur Heart J, 2018. 39(30): p. 2780-2792.
  3. McDonagh, T.A., et al.,Corrigendum to: 2021 ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure: Developed by the Task Force for the diagnosis and treatment of acute and chronic heart failure of the European Society of Cardiology (ESC) With the special contribution of the Heart Failure Association (HFA) of the ESC. Eur Heart J, 2021. 42(48): p. 4901.
  4. Heidenreich, P.A., et al.,2022 AHA/ACC/HFSA Guideline for the Management of Heart Failure: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation, 2022. 145(18): p. e895-e1032.
  5. Hébert, H.L., et al.,Cohort Profile: Genetics of Diabetes Audit and Research in Tayside Scotland (GoDARTS). Int J Epidemiol, 2018. 47(2): p. 380-381j.
  6. Jones, W.S., et al.,Comparative Effectiveness of Aspirin Dosing in Cardiovascular Disease. N Engl J Med, 2021. 384(21): p. 1981-1990.
  7. Hemingway, H., et al.,Big data from electronic health records for early and late translational cardiovascular research: challenges and potential. Eur Heart J, 2018. 39(16): p. 1481-1495.
  8. Bowton, E., et al.,Biobanks and electronic medical records: enabling cost-effective research. Sci Transl Med, 2014. 6(234): p. 234cm3.
  9. Allen, N.E., et al.,UK biobank data: come and get it. Sci Transl Med, 2014. 6(224): p. 224ed4.
  10. McKinstry, B., et al.,Cohort profile: the Scottish Research register SHARE. A register of people interested in research participation linked to NHS data sets. BMJ Open, 2017. 7(2): p. e013351.
  11. Nind, T., et al.,An extensible big data software architecture managing a research resource of real-world clinical radiology data linked to other health data from the whole Scottish population. Gigascience, 2020. 9(10).
  12. Tromp, J., Bauer, D., Claggett, B. L., Frost, M. W., Iversen, M., Prasad, N., Petrie, M. C., Larson, M. G., Ezekowitz, J. A., & Solomon, S. D. (2022). A formal validation of a deep learning-based automated workflow for the interpretation of the echocardiogram. 13(1).
  13. Tromp, J., Seekings, P. J., Hung, C.-L., Iversen, M. B., Frost, M. J., Ouwerkerk, W., Jiang, Z., Eisenhaber, F., Goh, R. S. M., Zhao, H., Huang, W., Ling, L.-H., Sim, D., Cozzone, P., Richards, A. M., Lee, H. K., Solomon, S. D., Lam, C. S. P., & Ezekowitz, J. A. (2021). Automated interpretation of systolic and diastolic function on the echocardiogram: a multicohort study.The Lancet Digital Health, 4(1).
  14. Hidenori Yaku, Komtebedde, J., Silvestry, F. E., & Sanjiv Jayendra Shah. (2024). Deep Learning-Based Automated Measurements Of Echocardiographic Estimators Of Invasive Pulmonary Capillary Wedge Pressure Perform Equally To Core Lab Measurements: Results From REDUCE LAP-HF II.Journal of the American College of Cardiology, 83(13), 316–316.
  15. OPERA – https://us2.ai/heart-failure-detection-faster-with-us2-ai-results-interchangeable-with-humans-the-opera-study/

MEDIA CONTACT

Company: Us2.ai

Name: Christine Gouillard

Email: christine@us2.ai

City: Singapore

Country: Singapore

Website: https://us2.ai/