Deep Learning Model For Early Detection Of Pneumonia: Implications For Public Health Screening
1
Department Of Pharmacy Practice, Dr. D. Y. Patil Institute Of Pharmaceutical Science And Research, Pimpri, India
Popul. Med. 2026;8(Supplement Supplement 1):
ABSTRACT
INTRODUCTION:
pneumonia remains one of the major causes of morbidity and mortality worldwide's most vulnerable group of cohorts in low and middle income nations. early diagnostic detection is critical for effective treatment; but in numerous resource-strained environments, there is low access in diagnostic services. the most common modality of pneumonia detection is chest x-ray imaging, which may be affected by latency and inter-observer variation in its interpretation. artificial intelligence (AI) can enlarge diagnostic capacity and strengthen public health systems due to quick and scalable image analysis. this study evaluates the quality of a deep-learning-implemented model to detect pneumonia in the chest x-ray images.
METHODS:
a retrospective examination was performed based on a publicly available dataset, containing about 5,000 chest x-ray films, whose annotation was pneumonia or normal. preprocessing and image augmentation were performed to increase model generalizability. through transfer learning, a convolutional-neural-network that relies on a pretrained densenet121 architecture was created. class imbalance was alleviated by weighted training, and a two-stage training program was used to minimize overfitting. the performance of models was measured on an independent test set and measures of accuracy, sensitivity, specificity, f1-score, and area under the receiver operating characteristic curve (AUC-ROC) were used. the process of threshold optimization was performed in order to align the results with the priorities of the public health screening.
RESULTS:
the model attained 83% accuracy and AUC-ROC of 0.93 at the optimal threshold of choice. pneumonia detection sensitivity was 96% and specificity to pick normal cases was 62%.
CONCLUSIONS:
the model was sensitive and well overall performed, which highlights the possibility of its application in the sphere of screening and decision-support tool in a local health-setting. implementing ai tools of this nature may lead to a higher success in detecting pneumonia earlier.
pneumonia remains one of the major causes of morbidity and mortality worldwide's most vulnerable group of cohorts in low and middle income nations. early diagnostic detection is critical for effective treatment; but in numerous resource-strained environments, there is low access in diagnostic services. the most common modality of pneumonia detection is chest x-ray imaging, which may be affected by latency and inter-observer variation in its interpretation. artificial intelligence (AI) can enlarge diagnostic capacity and strengthen public health systems due to quick and scalable image analysis. this study evaluates the quality of a deep-learning-implemented model to detect pneumonia in the chest x-ray images.
METHODS:
a retrospective examination was performed based on a publicly available dataset, containing about 5,000 chest x-ray films, whose annotation was pneumonia or normal. preprocessing and image augmentation were performed to increase model generalizability. through transfer learning, a convolutional-neural-network that relies on a pretrained densenet121 architecture was created. class imbalance was alleviated by weighted training, and a two-stage training program was used to minimize overfitting. the performance of models was measured on an independent test set and measures of accuracy, sensitivity, specificity, f1-score, and area under the receiver operating characteristic curve (AUC-ROC) were used. the process of threshold optimization was performed in order to align the results with the priorities of the public health screening.
RESULTS:
the model attained 83% accuracy and AUC-ROC of 0.93 at the optimal threshold of choice. pneumonia detection sensitivity was 96% and specificity to pick normal cases was 62%.
CONCLUSIONS:
the model was sensitive and well overall performed, which highlights the possibility of its application in the sphere of screening and decision-support tool in a local health-setting. implementing ai tools of this nature may lead to a higher success in detecting pneumonia earlier.
| eISSN: | 2654-1459 |
We process personal data collected when visiting the website. The function of obtaining information about users and their behavior is carried out by voluntarily entered information in forms and saving cookies in end devices. Data, including cookies, are used to provide services, improve the user experience and to analyze the traffic in accordance with the Privacy policy. Data are also collected and processed by Google Analytics tool (more).
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
