Effectiveness of artificial intelligence for lung disease screening in a municipal hospital

Background. To organize screening of the population for pulmonary tuberculosis, services based on the use of artificial intelligence technologies (AI services) have been developed and registered.

Aim: To evaluate diagnostic metrics and performance of the AI-service as medical decision support system within the framework of routine clinical practice at the scale of a municipal hospital.

Material and Methods. The index test was conducted using the software “Automated analysis program for digital chest X-ray/ fluorography images according to TU 62.01.29-001-96876180-2019” produced by LLC “PhthisisBiomed”.

Results. The index test of the AI service as a system for supporting medical decision-making showed high values of operational characteristics (sensitivity 96%, specificity 61%), significant savings in the time spent on forming conclusions, and high data transfer rate. The choice of the optimal separation point for screening is reasonably based on the metric of maximizing the predictive value of a negative result (sensitivity maximization). When comparing the diagnostic efficiency of AI-service solutions and physicians, it is shown that the area under the ROC curve of AI-service conclusions (0.91–0.93) is not inferior to that of qualified radiologists (0.78-0.91 according to the literature.

Discussion. The use of AI service allows to significantly save the time required to analyze one X-ray image, which is especially important for rapid diagnostics within the framework of screening programs. The use of AI service with high diagnostic efficiency expands the capabilities of radiologists and indicates a transition to a new level of quality of medical care. High speed data transfer allows for better coordination between medical staff and enables faster decision-making for patients.

Conclusions. Detection of pathological changes on radiographs of patients using AI-service has high diagnostic efficiency and can be used within the framework of population screening programs for lung diseases.

1. Afanasyeva E. N. Artificial Intelligence and Big Data in Healthcare: Applications and Legal Regulation. Juridical Science and Practice. 2020;16(3):40–49. (In Russ.). https://doi.org/10.25205/2542-0410-2020-16-3-40-49

2. Goldina TA, Burmistrov VA, Efimenko IV, Khoroshevskiy V.F. Artificial Intelligence in Healthcare: Real World Data and Patient Voice – Are We Ready for New Realities? Medical Technologies. Assessment and Choice. 2021;43(2):22–31. (In Russ.). https://doi.org/10.17116/medtech20214302122.

3. Alikperova N.V. Artificial intelligence in healthcare: risks and opportunities. Health of the megapolis. 2023;4(3):41–49. (In Russ.). https://doi.org/10.47619/2713-2617.zm.2023.v.4i3;41–49

4. Melendez J. Sánchez C.I. Philipsen R.H., Maduskar P., Dawson R., Theron G. et al. An automated tuberculosis screening strategy combining X-ray-based computer-aided detection and clinical information. Sci. Rep. 2016;29(6):25265. https://doi.org/10.1038/srep25265

5. Rahman M.T., Codlin A.J., Rahman M.M., Nahar A., Reja M., Islam T. et al. An evaluation of automated chest radiography reading software for tuberculosis screening among public- and private-sector patients. Eur. Respir. 2017:49. https://doi.org/10.1183/13993003.02159-2016

6. Lakhani P., Sundaram B. Deep learning at chest radiography: automated classification of pulmonary tuberculosis by using convolutional neural networks. Radiology. 2017;284(2):574–582. https://doi.org/10.1148/radiol.2017162326

7. Jaeger S., Juarez-Espinosa O.H., Candemir S., Poostchi M., Yang F., Kim L. et al. Detecting drug-resistant tuberculosis in chest radiographs. Int. J. Comput. Assist. Radiol. Surg. 2018;13(12):1915–1925. https://doi.org/10.1007/s11548-018-1857-9

8. Vajda S., Karargyris A., Jaeger S., Santosh Kc., Candemir S., Xue Zh. et al. Feature selection for automatic tuberculosis screening in frontal chest radiographs. J. Med. Syst. 2018;42(8):146. https://link.springer.com/article/10.1007/s10916-018-0991-9

9. Miroshnichenko S.I., Kovalenko Yu.N., Chernetsov V.B. Replacing fluorography with screening digital radiography. Polyclinic. 2016;6:19–22. (In Russ.).

10. Arzamasov K.M., Semenov S.S., Kokina D.Yu., Bobrovskaya T.M., Pavlov N.A., Kirpichev Yu.S. et al. Criteria for the applicability of computer vision for preventive studies using the example of X-ray and fluoroscopy of the organs of the chest. Medical physics. 2022;4:96:56. (In Russ.). https://doi.org/10.52775/1810-200X-2022-96-4-56-63

11. Gusev A., Morozov S., Lebedev G., Vladzimirsky A., Zinchenko V., Sharova D., et al. The development of artificial intelligence in healthcare in Russia. The reference library of intelligent systems. 2021:259–279. (In Russ.).

12. Gogoberidze Y.T., Klassen V.I., Natenzon M.Y., Prosvirkin I.A., Vladzimirsky A.V., Sharova D.E., et al. PhthisisBioMed artificial medical intelligence: software for automated analysis of digital chest x-ray/fluoro grams. Sovremennye tehnologii v medicine. 2023;15(4):5. (In Russ.). https://doi.org/10.17691/stm2023.15.4.01

13. Hwang E.J., Park S., Jin K., Kim J.I., Choi S.Y., Lee J.H. et al. Development and validation of a deep learning-based automated detection algorithm for major thoracic diseases on chest radiographs. JAMA Netw. Open. 2019;2(3):191095. https://doi.org/10.1001/jamanetworkopen.2019.1095