Identifying Medical Diagnoses and Treatable Diseases by Image-Based Deep Learning.

The implementation of clinical-decision support algorithms for medical imaging faces challenges with reliability and interpretability. Here, we establish a diagnostic tool based on a deep-learning framework for the screening of patients with common treatable blinding retinal diseases. Our framework utilizes transfer learning, which trains a neural network with a fraction of the data of conventional approaches. Applying this approach to a dataset of optical coherence tomography images, we demonstrate performance comparable to that of human experts in classifying age-related macular degeneration and diabetic macular edema. We also provide a more transparent and interpretable diagnosis by highlighting the regions recognized by the neural network. We further demonstrate the general applicability of our AI system for diagnosis of pediatric pneumonia using chest X-ray images. This tool may ultimately aid in expediting the diagnosis and referral of these treatable conditions, thereby facilitating earlier treatment, resulting in improved clinical outcomes. VIDEO ABSTRACT.

The SARS-CoV-2 spike L452R-E484Q variant in the Indian B.1.617 strain showed significant reduction in the neutralization activity of immune sera.

To assess the impact of the key non-synonymous amino acid substitutions in the RBD of the spike protein of SARS-CoV-2 variant B.1.617.1 (dominant variant identified in the current India outbreak) on the infectivity and neutralization activities of the immune sera, L452R and E484Q (L452R-E484Q variant), pseudotyped virus was constructed (with the D614G background). The impact on binding with the neutralizing antibodies was also assessed with an ELISA assay. Pseudotyped virus carrying a L452R-E484Q variant showed a comparable infectivity compared with D614G. However, there was a significant reduction in the neutralization activity of the immune sera from non-human primates vaccinated with a recombinant receptor binding domain (RBD) protein, convalescent patients, and healthy vaccinees vaccinated with an mRNA vaccine. In addition, there was a reduction in binding of L452R-E484Q-D614G protein to the antibodies of the immune sera from vaccinated non-human primates. These results highlight the interplay between infectivity and other biologic factors involved in the natural evolution of SARS-CoV-2. Reduced neutralization activities against the L452R-E484Q variant will have an impact on health authority planning and implications for the vaccination strategy/new vaccine development.

Mutation spectrum in GNAQ and GNA11 in Chinese uveal melanoma.

Uveal melanoma is the most common intraocular cancer in the adult eye. R183 and Q209 were found to be mutational hotspots in exon 4 and exon 5 of GNAQ and GNA11 in Caucasians. However, only a few studies have reported somatic mutations in GNAQ or GNA11 in uveal melanoma in Chinese. We extracted somatic DNA from paraffin-embedded biopsies of 63 Chinese uveal melanoma samples and sequenced the entire coding regions of exons 4 and 5 in GNAQ and GNA11. The results showed that 33% of Chinese uveal melanoma samples carried Q209 mutations while none had R183 mutation in GNAQ or GNA11. In addition, seven novel missense somatic mutations in GNAQ (Y192C, F194L, P170S, D236N, L232F, V230A, and M227I) and four novel missense somatic mutations in GNA11 (R166C, I200T, S225F, and V206M) were found in our study. The high mutation frequency of Q209 and the novel missense mutations detected in this study suggest that GNAQ and GNA11 are common targets for somatic mutations in Chinese uveal melanoma.

Evaluation and accurate diagnoses of pediatric diseases using artificial intelligence.

Artificial intelligence (AI)-based methods have emerged as powerful tools to transform medical care. Although machine learning classifiers (MLCs) have already demonstrated strong performance in image-based diagnoses, analysis of diverse and massive electronic health record (EHR) data remains challenging. Here, we show that MLCs can query EHRs in a manner similar to the hypothetico-deductive reasoning used by physicians and unearth associations that previous statistical methods have not found. Our model applies an automated natural language processing system using deep learning techniques to extract clinically relevant information from EHRs. In total, 101.6 million data points from 1,362,559 pediatric patient visits presenting to a major referral center were analyzed to train and validate the framework. Our model demonstrates high diagnostic accuracy across multiple organ systems and is comparable to experienced pediatricians in diagnosing common childhood diseases. Our study provides a proof of concept for implementing an AI-based system as a means to aid physicians in tackling large amounts of data, augmenting diagnostic evaluations, and to provide clinical decision support in cases of diagnostic uncertainty or complexity. Although this impact may be most evident in areas where healthcare providers are in relative shortage, the benefits of such an AI system are likely to be universal.