Human ; Writing ; Artificial Intelligence

In the last decade, artificial intelligence (AI) has been increasingly used in various fields of medicine. Recently, the advent of whole slide images (WSI) or digitized slides has paved the way for AI-based anatomic pathology. This paper set out to review the potential integration of AI algorithms in the workflow, and the utilization of AI in the practice of breast pathology.
Artificial Intelligence ; Breast Neoplasms

Twenty-six years earlier in their famous chess rematch, an IBM Supercomputer called Deep Blue defeated then-world chess champion Garry Kasparov: it was the first-ever chess match won by a machine, a much celebrated milestone in the field of Artificial Intelligence. Just last year, the World Association of Medical Editors released the “WAME Recommendations on Chatbots and Generative Artificial Intelligence in Relation to Scholarly Publications,” a recognition of not just the expanding applications of AI in scholarly publishing but more so of the accompanying emergence of concerns on authenticity and accuracy. In recognition of this relevant topic, our Vice Editor in Chief, Dr. Cecile Jimeno, provided a well-attended and interesting talk during the last ASEAN Federation of Endocrine Society Convention in Thailand on the “Emerging Issues on the Use of Artificial Intelligence for Scientific Publications.”
Artificial Intelligence

Background: Proper assessment and efficient diagnosis of central nervous system anomalies is essential in antenatal surveillance of pregnant patients. These anomalies are usually associated with genetic syndromes or severe malformations requiring timely intervention and antenatal counseling of the expectant couple. Objective: The study aims to evaluate the agreement of cranial biometric measurements and to determine if there is a significant difference in the time needed to complete the evaluation using standard 2D and semi-automated 5D ultrasound. Methods: An analytical cross-sectional study was employed on 93 women who underwent pelvic ultrasound scans from August to October 2022 in a tertiary hospital. Basic biometric fetal central nervous system (CNS) measurements were acquired using 2D ultrasound followed by 5D CNS ultrasound. Bland-Altman plots were used to evaluate the agreement of the measurements obtained. The difference in the time to completion was determined using independent t-test. Results and Conclusions Our study found that 5D CNS ultrasound measurements showed 96.8% agreement with 2D ultrasound in 90 out of 93 fetuses. The 5D CNS ultrasound takes a shorter time of 90 seconds (s) to completion in comparison to 99 s using the 2D method (p=0.076). Upon stratification of the study population per trimester, in the second trimester, it took 76 s with 5D CNS vs 89 s with 2D, resulting to a statistically significant 13-second difference (p=0.044). In the third trimester, 5D CNS took 105 s vs 108 s with 2D (p=0.614). The time to completion of the scan using this technology is faster when used for second trimester pregnancies but could be affected by fetal-dependent and operator-dependent factors. Therefore, application of this new technology has the potential to improve workflow efficiency after the necessary training on 3D sonography and 5D CNS ultrasound software.
Artificial Intelligence

Humans ; Artificial Intelligence ; Algorithms ; Brain ; Head

Cardiovascular risk assessment is a basic tenet of the prevention of cardiovascular disease. Conventional risk assessment models require measurements of blood pressure, blood lipids, and other health-related information prior to assessment of risk via regression models. Compared with traditional approaches, fundus photograph-based cardiovascular risk assessment using artificial intelligence (AI) technology is novel, and has the advantages of immediacy, non-invasiveness, easy performance, and low cost. The Health Risk Assessment and Control Committee of the Chinese Preventive Medicine Association, in collaboration with the Chinese Society of Cardiology and the Society of Health Examination, invited multi-disciplinary experts to form a panel to develop the present consensus, which includes relevant theories, progress in research, and requirements for AI model development, as well as applicable scenarios, applicable subjects, assessment processes, and other issues associated with applying AI technology to assess cardiovascular risk based on fundus photographs. A consensus was reached after multiple careful discussions on the relevant research, and the needs of the health management industry in China and abroad, in order to guide the development and promotion of this new technology.
Humans ; Cardiovascular Diseases/prevention & control* ; Artificial Intelligence ; Consensus ; Risk Factors ; Heart Disease Risk Factors

OBJECTIVES

This study evaluated the accuracy of an artificial intelligence (AI) model in identifying retinal lesions, validated its performance on a Filipino population dataset, and evaluated the impact of dataset diversity on AI analysis accuracy.

This cross-sectional, analytical, institutional study analyzed standardized macula-centered fundus photos taken with the Zeiss Visucam®. The AI model’s output was compared with manual readings by trained retina specialists.

A total of 215 eyes from 109 patients were included in the study. Human graders identified 109 eyes (50.7%) with retinal abnormalities. The AI model demonstrated an overall accuracy of 73.0% (95% CI 66.6% – 78.8%) in detecting abnormal retinas, with a sensitivity of 54.1% (95% CI 44.3% – 63.7%) and specificity of 92.5% (95% CI 85.7% – 96.7%).

The availability and sources of AI training datasets can introduce biases into AI algorithms. In our dataset, racial differences in retinal morphology, such as differences in retinal pigmentation, affected the accuracy of AI image-based analysis. More diverse datasets and external validation on different populations are needed to mitigate these biases.

OBJECTIVE

To develop and evaluate the diagnostic performance of an automated machine learning (AutoML) model for the detection of referable diabetic retinopathy (refDR) in ultrawide field (UWF) retinal images from local Philippine retinal image datasets.

A Google AutoML Vision model was trained using 2000 UWF images with a 50/50 ratio of refDR/non-refDR. Images were labeled according to the Early Treatment Diabetic Retinopathy Study (ETDRS) severity grading. RefDR was defined as moderate nonproliferative DR or worse. The dataset was split with 80% for training, 10% for validation, and 10% for testing. Two sets of published UWF image sets were used for external validation. Sensitivity and specificity were calculated in accordance with United States Food and Drug Administration (US FDA) performance requirements of 0.85 and 0.825, respectively.

The area under the precision-recall curve was 0.998. External validation against two datasets showed a sensitivity/specificity of 0.88/0.83 (95% CI 0.80-0.94/0.74-0.89) and 0.83/0.80 (95% CI 0.74-0.89/0.72-0.86), respectively. Positive and negative predictive values were 0.81/0.89 (95% CI 0.73-0.89/0.82-0.94) and 0.75/0.86 (95% CI 0.66-0.83/0.79-0.91), respectively.

The pilot performance of the custom AutoML model constructed using local Philippine data approaches US FDA requirements for the diagnosis of referable DR. The ease of use and intuitiveness of the platform, combined with its performance, support the potential of no-code AI in the detection of refDR.

Humans ; Artificial Intelligence ; Lung Neoplasms/pathology* ; Mediastinum/pathology* ; Lymph Nodes/pathology* ; Neoplasm Staging ; Lymph Node Excision

Humans ; Artificial Intelligence ; Publications