Role of Artificial Intelligence in Identifying Vital Biomarkers with Greater Precision in Emergency Departments During Emerging Pandemics

International Journal of Molecular Sciences, Journal Year: 2025, Volume and Issue: 26(2), P. 722 - 722

Published: Jan. 16, 2025

The COVID-19 pandemic has accelerated advances in molecular biology and virology, enabling the identification of key biomarkers to differentiate between severe mild cases. Furthermore, use artificial intelligence (AI) machine learning (ML) analyze large datasets been crucial for rapidly identifying relevant disease prognosis, including COVID-19. This approach enhances diagnostics emergency settings, allowing more accurate efficient patient management. study demonstrates how algorithms departments can identify vital prognosis an emerging using as example by analyzing clinical, epidemiological, analytical, radiological data. All consecutively admitted patients were included, than 89 variables processed Random Forest (RF) algorithm. RF model achieved highest balanced accuracy at 92.61%. most predictive mortality included procalcitonin (PCT), lactate dehydrogenase (LDH), C-reactive protein (CRP). Additionally, system highlighted significance interstitial infiltrates chest X-rays D-dimer levels. Our results demonstrate that is critical diseases, accelerating data analysis, optimizing personalized treatment, emphasizing importance PCT LDH high-risk patients.

Language: Английский

---

Accuracy of the AI-Based Smart Scope® Test as a Point-of-Care Screening and Triage Tool Compared to Colposcopy: A Pilot Study

Cureus, Journal Year: 2025, Volume and Issue: unknown

Published: March 26, 2025

Objectives The primary objective of this study was to compare the screening accuracy AI assessment with colposcopy. Secondary objectives included comparing triaging and colposcopy assessments against histopathology. Methodology This prospective, single-arm test conducted at obstetrics gynecology department Bharati Vidyapeeth (Deemed be University) Medical College in Pune, India. sexually active, nonpregnant women aged 25-65 years visiting OPD for per-speculum examination. Women a clinically unhealthy cervix detected during examination were counseled, those who provided consent enrolled. Patients history prior cervical cancer treatment or hysterectomy excluded. A total 130 Each participant underwent colposcopy, Smart Scope®-AI (SS-AI) assisted visual inspection acetic acid (VIA), Lugol's iodine same visit. Positive findings from any led biopsy, samples sent histopathological analysis. Results Of enrolled, 30 referred biopsy. Histopathology results obtained 18 consenting women. Using as reference standard (N = 130), SS-AI 76.53%. When compared histopathology 18) gold standard, 63.67% 83.33%, respectively. sensitivity specificity both while had 83.33% 50%. Likelihood ratios superior These suggest that SS-AI-assisted test, digital VIA accurately detects positive negative lesions. Conclusions system demonstrated comparable effectiveness has potential used point-of-care tool healthcare centers lacking equipment purposes.

Language: Английский

---

AI-Based Identification Method for Cervical Transformation Zone Within Digital Colposcopy: Development and Multicenter Validation Study

JMIR Cancer, Journal Year: 2025, Volume and Issue: 11, P. e69672 - e69672

Published: March 31, 2025

Background In low- and middle-income countries, cervical cancer remains a leading cause of death morbidity for women. Early detection treatment precancerous lesions are critical in prevention, colposcopy is primary diagnostic tool identifying guiding biopsies. The transformation zone (TZ) where stratified squamous epithelium develops from the metaplasia simple columnar most common site lesions. However, inexperienced colposcopists may find it challenging to accurately identify type location TZ during examination. Objective This study aims present an artificial intelligence (AI) method enhance examination evaluate its potential clinical application. Methods retrospectively collected data 3616 women who underwent at 6 tertiary hospitals China between 2019 2021. A dataset 4 was model conduction. An independent other 2 geographic validate performance. There no overlap training validation datasets. Anonymized digital records, including each image, baseline characteristics, colposcopic findings, pathological outcomes, were collected. classification proposed as lightweight neural network with multiscale feature enhancement capabilities designed classify 3 types TZ. pretrained FastSAM first implemented new squamocolumnar junction segmenting Overall accuracy, average precision, recall evaluated segmentation models. performance on external assessed by sensitivity specificity. Results optimal performed 83.97% accuracy test set, which achieved precision 91.84%, 89.06%, 95.62% 1, 2, 3, respectively. mean 0.78 0.75, demonstrated outstanding predicting TZ, achieving 95% CIs TZ1, TZ2, TZ3 (0.74-0.81), 0.81 (0.78-0.82), 0.8 (0.74-0.87), respectively, specificity 0.94 (0.92-0.96), 0.83 (0.81-0.86), 0.91 (0.89-0.92), based comprehensive 1335 cases hospitals. Conclusions Our AI-based identification system classified TZs delineated their multicenter, colposcopic, high-resolution images. findings this have shown predict specific regions accurately. It developed valuable assistant encourage precise practice.

Language: Английский

---

Real time mobile AI-assisted cervicography interpretation system

Informatics in Medicine Unlocked, Journal Year: 2023, Volume and Issue: 42, P. 101360 - 101360

Published: Jan. 1, 2023

Cervicography visual inspection after acetic acid application (VIA) has been recognized as an alternative early screening in resource-limited settings, such Indonesia. However, the accuracy of VIA results primarily relies on examiner's expertise, and due to inadequate comprehensive training healthcare workers, is diminishing. Our primary goal was develop a real time mobile AI-assisted cervicography interpretation system empowered by lightweight model promptly autonomously determine precise results. custom dataset comprises substantial collection 702 subjects from Dr Mohammad Hoesin General Hospital, Indonesia which were classified into two conditions: 418 with abnormal cervixes 302 control. We conducted experiments: one focused detection region interest (RoI) cervix, other segmentation precancerous lesions. In this study, we utilize object approach using combined You Only Look Once (YOLO) framework. As result, proposed achieves exceptional mean average precision (mAP) 99% for RoI cervix detection, while lesions mAP 73% intersection over union score 40%. Furthermore, showcases inference 10.4 ms, reflecting its efficiency processing images generating swiftly. also assessed oncologist consultants, indicated satisfactory agreement Kappa value 0.838. The high signifies level between model's predictions assessments made consultants. This further validates effectiveness lesion highlights potential utility clinical settings.

Language: Английский

---

Performance of artificial intelligence for diagnosing cervical intraepithelial neoplasia and cervical cancer: a systematic review and meta-analysis

EClinicalMedicine, Journal Year: 2024, Volume and Issue: 80, P. 102992 - 102992

Published: Dec. 28, 2024

Language: Английский

---

Enhancing the Accuracy of Lymph-Node-Metastasis Prediction in Gynecologic Malignancies Using Multimodal Federated Learning: Integrating CT, MRI, and PET/CT

Cancers, Journal Year: 2023, Volume and Issue: 15(21), P. 5281 - 5281

Published: Nov. 3, 2023

Gynecological malignancies, particularly lymph node metastasis, have presented a diagnostic challenge, even with traditional imaging techniques such as CT, MRI, and PET/CT. This study was conceived to explore and, subsequently, bridge this gap through more holistic innovative approach. By developing comprehensive framework that integrates both non-image data detailed MRI image analyses, harnessed the capabilities of multimodal federated-learning model. Employing composite neural network within environment, adeptly merged diverse sources enhance prediction accuracy. further complemented by sophisticated deep convolutional an enhanced U-NET architecture for meticulous processing. Traditional yielded sensitivities ranging from 32.63% 57.69%. In contrast, model, without incorporating data, achieved impressive sensitivity approximately 0.9231, which soared 0.9412 integration data. Such advancements underscore significant potential approach, suggesting federated learning, especially when combined assessment can revolutionize lymph-node-metastasis detection in gynecological malignancies. paves way precise patient care, potentially transforming current paradigm resulting improved outcomes.

Language: Английский

---