Few shot learning for phenotype-driven diagnosis of patients with rare genetic diseases

Abstract

There are more than 7,000 rare diseases, some of which affect 3,500 or fewer patients in the US. Due to clinicians' limited experience with such diseases and the considerable heterogeneity of their clinical presentations, many patients with rare genetic diseases remain undiagnosed. While artificial intelligence has demonstrated success in assisting diagnosis, its success is usually contingent on the availability of large labeled datasets. Here, we present SHEPHERD, a deep learning approach for multi-faceted rare disease diagnosis. SHEPHERD is guided by existing knowledge of diseases, phenotypes, and genes to learn novel connections between a patient's clinico-genetic information and phenotype and gene relationships. We train SHEPHERD exclusively on simulated patients and evaluate on a cohort of 465 patients representing 299 diseases (79% of genes and 83% of diseases are represented in only a single patient) in the Undiagnosed Diseases Network. SHEPHERD excels at several diagnostic facets: performing causal gene discovery (causal genes are predicted at rank = 3.52 on average), retrieving "patients-like-me" with the same gene or disease, and providing interpretable characterizations of novel disease presentations. SHEPHERD demonstrates the potential of artificial intelligence to accelerate the diagnosis of rare disease patients and has implications for the use of deep learning on medical datasets with very few labels.

A foundation model for clinician-centered drug repurposing

“…TXGNN has zero-shot predicting ability of therapeutic use for diseases without existing treatments and limited molecular understanding. This opens up new avenues for machine learn-14 ing to identify therapeutic opportunities for diseases that are challenging to model using existing strategies as well as neglected and rare diseases 35 in urgent need of therapeutic innovation 36,37 .…”

“…This overlooks the vast array of disorders-92% of the 17,080 diseases we analyzed-lacking such pre-existing indications and in-depth molecular characterization. The clinical imperative thus leans heavily on addressing the needs of these lesserknown disorders, many of which fall into the category of complex, neglected, or rare diseases [45][46][47] . This has culminated in what we refer to as zero-shot drug repurposing.…”

“…This overlooks the vast array of diseases-92% of the 17,080 diseases we analyzed-lacking such pre-existing indications and known molecular target interactions. Addressing the needs of these diseases, many of which are complex, neglected, or rare, is a top clinical priority [55][56][57] . We define this challenge as zero-shot drug repurposing.…”

“…First, TXGNN shows consistently strong performance for neglected diseases in all settings. Across six settings, TXGNN gains up to 49.2% (average gain = 33.9%) in accuracy on predicting indications and up to 35.1% (average gain = 26.8%) in accuracy on predicting contraindications compared to a state-ofthe-art graph neural network. Second, we evaluated TXGNN for diseases not encountered during machine learning model training.…”

The Feasibility and Performance of Total Hip Replacement Prediction Deep Learning Algorithm with Real World Data

“…Several algorithms were developed to detect skeletal abnormalities in plain films [ 26 , 27 , 28 ]. From acute traumatic fracture, congenital disorder, oncologic lesion to degenerative disease, there was evidence to support the DL algorithm that can help the clinical doctors to detect the lesion accurately and efficiently [ 29 , 30 , 31 , 32 ]. For degenerative disorders, there are also several references that provide significant evidence which DL can offer proper support for clinical doctors.…”