Automated Machine Learning versus Expert-Designed Models in Ocular Toxoplasmosis: Detection and Lesion Localization Using Fundus Images.

PURPOSE: Automated machine learning (AutoML) allows clinicians without coding experience to build their own deep learning (DL) models. This study assesses the performance of AutoML in detecting and localizing ocular toxoplasmosis (OT) lesions in fundus images and compares it to expert-designed models. METHODS: Ophthalmology trainees without coding experience designed AutoML models using 304 labelled fundus images. We designed a binary model to differentiate OT from normal and an object detection model to visually identify OT lesions. RESULTS: The AutoML model had an area under the precision-recall curve (AuPRC) of 0.945, sensitivity of 100%, specificity of 83% and accuracy of 93.5% (vs. 94%, 86% and 91% for the bespoke models). The AutoML object detection model had an AuPRC of 0.600 with a precision of 93.3% and recall of 56%. Using a diversified external validation dataset, our model correctly labeled 15 normal fundus images (100%) and 15 OT fundus images (100%), with a mean confidence score of 0.965 and 0.963, respectively. CONCLUSION: AutoML models created by ophthalmologists without coding experience were comparable or better than expert-designed bespoke models trained on the same dataset. By creatively using AutoML to identify OT lesions on fundus images, our approach brings the whole spectrum of DL model design into the hands of clinicians.

Assessing the medical reasoning skills of GPT-4 in complex ophthalmology cases.

BACKGROUND/AIMS: This study assesses the proficiency of Generative Pre-trained Transformer (GPT)-4 in answering questions about complex clinical ophthalmology cases. METHODS: We tested GPT-4 on 422 Journal of the American Medical Association Ophthalmology Clinical Challenges, and prompted the model to determine the diagnosis (open-ended question) and identify the next-step (multiple-choice question). We generated responses using two zero-shot prompting strategies, including zero-shot plan-and-solve+ (PS+), to improve the reasoning of the model. We compared the best-performing model to human graders in a benchmarking effort. RESULTS: Using PS+ prompting, GPT-4 achieved mean accuracies of 48.0% (95% CI (43.1% to 52.9%)) and 63.0% (95% CI (58.2% to 67.6%)) in diagnosis and next step, respectively. Next-step accuracy did not significantly differ by subspecialty (p=0.44). However, diagnostic accuracy in pathology and tumours was significantly higher than in uveitis (p=0.027). When the diagnosis was accurate, 75.2% (95% CI (68.6% to 80.9%)) of the next steps were correct. Conversely, when the diagnosis was incorrect, 50.2% (95% CI (43.8% to 56.6%)) of the next steps were accurate. The next step was three times more likely to be accurate when the initial diagnosis was correct (p<0.001). No significant differences were observed in diagnostic accuracy and decision-making between board-certified ophthalmologists and GPT-4. Among trainees, senior residents outperformed GPT-4 in diagnostic accuracy (p≤0.001 and 0.049) and in accuracy of next step (p=0.002 and 0.020). CONCLUSION: Improved prompting enhances GPT-4's performance in complex clinical situations, although it does not surpass ophthalmology trainees in our context. Specialised large language models hold promise for future assistance in medical decision-making and diagnosis.

Posterior Corneal Astigmatism Does Not Influence Manifest-Treated Topography-guided LASIK Outcomes.

PURPOSE: To investigate whether the magnitude of posterior corneal astigmatism (PCA) impacts refractive and visual outcomes of primary topography-guided laser in situ keratomileusis (LASIK) and to provide guidance on treating eyes with high PCA. METHODS: Comparative retrospective analysis of 4,541 consecutive eyes treated with Contoura (Alcon Laboratories, Inc) on the manifest refractive astigmatism. Standard outcomes of the 1,514 eyes with the lowest PCA (first tercile; low PCA group) were compared to the 1,514 eyes with the highest PCA (last tercile; high PCA group). Pearson correlation coefficient was used to assess relationships between variables. RESULTS: Preoperatively, 20.9% of eyes presented with PCA of 0.50 diopters (D) or greater. The mean PCA was 0.18 ± 0.07 D in eyes with low PCA, and 0.50 ± 0.11 D in eyes with high PCA. An equivalent number of eyes achieved a cumulative postoperative unilateral uncorrected distance visual acuity of 20/20 in both the low PCA and high PCA groups (95.3% vs 94.7%; P = .4489). The efficacy index of both low and high PCA eyes was identical (0.99 ± 0.06 vs 0.99 ± 0.08; P = .3192), as was the safety index (1.00 ± 0.02 vs 1.00 ± 0.03; P = .0110). The magnitude of preoperative PCA was weakly correlated with postoperative refractive astigmatism (R = 0.1323), but not with postoperative defocus equivalent (R = -0.0414) or spherical equivalent (R = -0.0128). CONCLUSIONS: PCA does not negatively impact the outcomes of topography-guided LASIK targeting the manifest refraction, having identical accuracy, efficacy, and safety in eyes with both low and high PCA. There is no scientific basis to measure and consider PCA in topography-guided LASIK planning software or nomograms if the excimer laser treatment input targets the manifest refraction. [J Refract Surg. 2022;38(12):780-790.].

Portrait of DNA methylated genes predictive of poor prognosis in head and neck cancer and the implication for targeted therapy.

In addition to chronic infection with human papilloma virus (HPV) and exposure to environmental carcinogens, genetic and epigenetic factors act as major risk factors for head and neck cancer (HNC) development and progression. Here, we conducted a systematic review in order to assess whether DNA hypermethylated genes are predictive of high risk of developing HNC and/or impact on survival and outcomes in non-HPV/non-tobacco/non-alcohol associated HNC. We identified 85 studies covering 32,187 subjects where the relationship between DNA methylation, risk factors and survival outcomes were addressed. Changes in DNA hypermethylation were identified for 120 genes. Interactome analysis revealed enrichment in complex regulatory pathways that coordinate cell cycle progression (CCNA1, SFN, ATM, GADD45A, CDK2NA, TP53, RB1 and RASSF1). However, not all these genes showed significant statistical association with alcohol consumption, tobacco and/or HPV infection in the multivariate analysis. Genes with the most robust HNC risk association included TIMP3, DCC, DAPK, CDH1, CCNA1, MGMT, P16, MINT31, CD44, RARß. From these candidates, we further validated CD44 at translational level in an independent cohort of 100 patients with tongue cancer followed-up beyond 10 years. CD44 expression was associated with high-risk of tumor recurrence and metastasis (P = 0.01) in HPV-cases. In summary, genes regulated by methylation play a modulatory function in HNC susceptibility and it represent a critical therapeutic target to manage patients with advanced disease.