PubCaseFinder: A Case-Report-Based, Phenotype-Driven Differential-Diagnosis System for Rare Diseases

Objectivizing issues in the diagnosis of complex rare diseases: lessons learned from testing existing diagnosis support systems on ciliopathies

BACKGROUND

There are approximately 8,000 different rare diseases that affect roughly 400 million people worldwide. Many of them suffer from delayed diagnosis. Ciliopathies are rare monogenic disorders characterized by a significant phenotypic and genetic heterogeneity that raises an important challenge for clinical diagnosis. Diagnosis support systems (DSS) applied to electronic health record (EHR) data may help identify undiagnosed patients, which is of paramount importance to improve patients' care. Our objective was to evaluate three online-accessible rare disease DSSs using phenotypes derived from EHRs for the diagnosis of ciliopathies.

METHODS

Two datasets of ciliopathy cases, either proven or suspected, and two datasets of controls were used to evaluate the DSSs. Patient phenotypes were automatically extracted from their EHRs and converted to Human Phenotype Ontology terms. We tested the ability of the DSSs to diagnose cases in contrast to controls based on Orphanet ontology.

RESULTS

A total of 79 cases and 38 controls were selected. Performances of the DSSs on ciliopathy real world data (best DSS with area under the ROC curve = 0.72) were not as good as published performances on the test set used in the DSS development phase. None of these systems obtained results which could be described as "expert-level". Patients with multisystemic symptoms were generally easier to diagnose than patients with isolated symptoms. Diseases easily confused with ciliopathy generally affected multiple organs and had overlapping phenotypes. Four challenges need to be considered to improve the performances: to make the DSSs interoperable with EHR systems, to validate the performances in real-life settings, to deal with data quality, and to leverage methods and resources for rare and complex diseases.

CONCLUSION

Our study provides insights into the complexities of diagnosing highly heterogenous rare diseases and offers lessons derived from evaluation existing DSSs in real-world settings. These insights are not only beneficial for ciliopathy diagnosis but also hold relevance for the enhancement of DSS for various complex rare disorders, by guiding the development of more clinically relevant rare disease DSSs, that could support early diagnosis and finally make more patients eligible for treatment.

Performance and clinical utility of a new supervised machine-learning pipeline in detecting rare ciliopathy patients based on deep phenotyping from electronic health records and semantic similarity

BACKGROUND

Rare diseases affect approximately 400 million people worldwide. Many of them suffer from delayed diagnosis. Among them, NPHP1-related renal ciliopathies need to be diagnosed as early as possible as potential treatments have been recently investigated with promising results. Our objective was to develop a supervised machine learning pipeline for the detection of NPHP1 ciliopathy patients from a large number of nephrology patients using electronic health records (EHRs).

METHODS AND RESULTS

We designed a pipeline combining a phenotyping module re-using unstructured EHR data, a semantic similarity module to address the phenotype dependence, a feature selection step to deal with high dimensionality, an undersampling step to address the class imbalance, and a classification step with multiple train-test split for the small number of rare cases. The pipeline was applied to thirty NPHP1 patients and 7231 controls and achieved good performances (sensitivity 86% with specificity 90%). A qualitative review of the EHRs of 40 misclassified controls showed that 25% had phenotypes belonging to the ciliopathy spectrum, which demonstrates the ability of our system to detect patients with similar conditions.

CONCLUSIONS

Our pipeline reached very encouraging performance scores for pre-diagnosing ciliopathy patients. The identified patients could then undergo genetic testing. The same data-driven approach can be adapted to other rare diseases facing underdiagnosis challenges.

Phenotypic similarity-based approach for variant prioritization for unsolved rare disease: a preliminary methodological report

Rare diseases (RD) have a prevalence of not more than 1/2000 persons in the European population, and are characterised by the difficulty experienced in obtaining a correct and timely diagnosis. According to Orphanet, 72.5% of RD have a genetic origin although 35% of them do not yet have an identified causative gene. A significant proportion of patients suspected to have a genetic RD receive an inconclusive exome/genome sequencing. Working towards the International Rare Diseases Research Consortium (IRDiRC)'s goal for 2027 to ensure that all people living with a RD receive a diagnosis within one year of coming to medical attention, the Solve-RD project aims to identify the molecular causes underlying undiagnosed RD. As part of this strategy, we developed a phenotypic similarity-based variant prioritization methodology comparing submitted cases with other submitted cases and with known RD in Orphanet. Three complementary approaches based on phenotypic similarity calculations using the Human Phenotype Ontology (HPO), the Orphanet Rare Diseases Ontology (ORDO) and the HPO-ORDO Ontological Module (HOOM) were developed; genomic data reanalysis was performed by the RD-Connect Genome-Phenome Analysis Platform (GPAP). The methodology was tested in 4 exemplary cases discussed with experts from European Reference Networks. Variants of interest (pathogenic or likely pathogenic) were detected in 8.8% of the 725 cases clustered by similarity calculations. Diagnostic hypotheses were validated in 42.1% of them and needed further exploration in another 10.9%. Based on the promising results, we are devising an automated standardized phenotypic-based re-analysis pipeline to be applied to the entire unsolved cases cohort.

OARD: Open annotations for rare diseases and their phenotypes based on real-world data

Diagnosis for rare genetic diseases often relies on phenotype-driven methods, which hinge on the accuracy and completeness of the rare disease phenotypes in the underlying annotation knowledgebase. Existing knowledgebases are often manually curated with additional annotations found in published case reports. Despite their potential, real-world data such as electronic health records (EHRs) have not been fully exploited to derive rare disease annotations. Here, we present open annotation for rare diseases (OARD), a real-world-data-derived resource with annotation for rare-disease-related phenotypes. This resource is derived from the EHRs of two academic health institutions containing more than 10 million individuals spanning wide age ranges and different disease subgroups. By leveraging ontology mapping and advanced natural-language-processing (NLP) methods, OARD automatically and efficiently extracts concepts for both rare diseases and their phenotypic traits from billing codes and lab tests as well as over 100 million clinical narratives. The rare disease prevalence derived by OARD is highly correlated with those annotated in the original rare disease knowledgebase. By performing association analysis, we identified more than 1 million novel disease-phenotype association pairs that were previously missed by human annotation, and >60% were confirmed true associations via manual review of a list of sampled pairs. Compared to the manual curated annotation, OARD is 100% data driven and its pipeline can be shared across different institutions. By supporting privacy-preserving sharing of aggregated summary statistics, such as term frequencies and disease-phenotype associations, it fills an important gap to facilitate data-driven research in the rare disease community.

Seven years since the launch of the Matchmaker Exchange: The evolution of genomic matchmaking

The Matchmaker Exchange (MME) was launched in 2015 to provide a robust mechanism to discover novel disease-gene relationships. It operates as a federated network connecting databases holding relevant data using a common application programming interface, where two or more users are looking for a match for the same gene (two-sided matchmaking). Seven years from its launch, it is clear that the MME is making outstanding contributions to understanding the morbid anatomy of the genome. The number of unique genes present across the MME has steadily increased over time; there are currently >13,520 unique genes (~68% of all protein-coding genes) connected across the MME's eight genomic matchmaking nodes, GeneMatcher, DECIPHER, PhenomeCentral, MyGene2, seqr, Initiative on Rare and Undiagnosed Disease, PatientMatcher, and the RD-Connect Genome-Phenome Analysis Platform. The collective data set accessible across the MME currently includes more than 120,000 cases from over 12,000 contributors in 98 countries. The discovery of potential new disease-gene relationships is happening daily and international collaborative teams are moving these advances forward to publication, now numbering well over 500. Expansion of data sharing into routine clinical practice by clinicians, genetic counselors, and clinical laboratories has ensured access to discovery for even more individuals with undiagnosed rare genetic diseases. Tens of thousands of patients and their family members have been directly or indirectly impacted by the discoveries facilitated by two-sided genomic matchmaking. MME supports further connections to the literature (PubCaseFinder) and to human and model organism resources (Monarch Initiative) and scientists (ModelMatcher). Efforts are now underway to explore additional approaches to matchmaking at the gene or variant level where there is only one querier (one-sided matchmaking). Genomic matchmaking has proven its utility over the past 7 years and will continue to facilitate discoveries in the years to come.

Methods to Improve Molecular Diagnosis in Genomic Cold Cases in Pediatric Neurology

During the last decade, genetic testing has emerged as an important etiological diagnostic tool for Mendelian diseases, including pediatric neurological conditions. A genetic diagnosis has a considerable impact on disease management and treatment; however, many cases remain undiagnosed after applying standard diagnostic sequencing techniques. This review discusses various methods to improve the molecular diagnostic rates in these genomic cold cases. We discuss extended analysis methods to consider, non-Mendelian inheritance models, mosaicism, dual/multiple diagnoses, periodic re-analysis, artificial intelligence tools, and deep phenotyping, in addition to integrating various omics methods to improve variant prioritization. Last, novel genomic technologies, including long-read sequencing, artificial long-read sequencing, and optical genome mapping are discussed. In conclusion, a more comprehensive molecular analysis and a timely re-analysis of unsolved cases are imperative to improve diagnostic rates. In addition, our current understanding of the human genome is still limited due to restrictions in technologies. Novel technologies are now available that improve upon some of these limitations and can capture all human genomic variation more accurately. Last, we recommend a more routine implementation of high molecular weight DNA extraction methods that is coherent with the ability to use and/or optimally benefit from these novel genomic methods.

Advances in the development of PubCaseFinder, including the new application programming interface and matching algorithm

Over 10,000 rare genetic diseases have been identified, and millions of newborns are affected by severe rare genetic diseases each year. A variety of Human Phenotype Ontology (HPO)-based clinical decision support systems (CDSS) and patient repositories have been developed to support clinicians in diagnosing patients with suspected rare genetic diseases. In September 2017, we released PubCaseFinder (https://pubcasefinder.dbcls.jp), a web-based CDSS that provides ranked lists of genetic and rare diseases using HPO-based phenotypic similarities, where top-listed diseases represent the most likely differential diagnosis. We also developed a Matchmaker Exchange (MME) application programming interface (API) to query PubCaseFinder, which has been adopted by several patient repositories. In this paper, we describe notable updates regarding PubCaseFinder, the GeneYenta matching algorithm implemented in PubCaseFinder, and the PubCaseFinder API. The updated GeneYenta matching algorithm improves the performance of the CDSS automated differential diagnosis function. Moreover, the updated PubCaseFinder and new API empower patient repositories participating in MME and medical professionals to actively use HPO-based resources. This article is protected by copyright. All rights reserved.

PhenoApt leverages clinical expertise to prioritize candidate genes via machine learning

In recent years, exome sequencing (ES) has shown great utility in the diagnoses of Mendelian disorders. However, after rigorous filtering, a typical ES analysis still involves the interpretation of hundreds of variants, which greatly hinders the rapid identification of causative genes. Since the interpretations of ES data require comprehensive clinical analyses, taking clinical expertise into consideration can speed the molecular diagnoses of Mendelian disorders. To leverage clinical expertise to prioritize candidate genes, we developed PhenoApt, a phenotype-driven gene prioritization tool that allows users to assign a customized weight to each phenotype, via a machine-learning algorithm. Using the ability to rank causative genes in top-10 lists as an evaluation metric, baseline analysis demonstrated that PhenoApt outperformed previous phenotype-driven gene prioritization tools by a relative increase of 22.7%-140.0% in three independent, real-world, multi-center cohorts (cohort 1, n = 185; cohort 2, n = 784; and cohort 3, n = 208). Additional trials showed that, by adding weights to clinical indications, which should be explained by the causative gene, PhenoApt performance was improved by a relative increase of 37.3% in cohort 2 (n = 471) and 21.4% in cohort 3 (n = 208). Moreover, PhenoApt could assign an intrinsic weight to each phenotype based on the likelihood of its being a Mendelian trait using term frequency-inverse document frequency techniques. When clinical indications were assigned with intrinsic weights, PhenoApt performance was improved by a relative increase of 23.7% in cohort 2 and 15.5% in cohort 3. For the integration of PhenoApt into clinical practice, we developed a user-friendly website and a command-line tool.

2022: a pivotal year for diagnosis and treatment of rare genetic diseases

The start of 2022 is an inflection point in the development of diagnostics and treatments for rare genetic diseases in prenatal, pediatric, and adult individuals-the theme of this special issue. Here I briefly review recent developments in two pivotal aspects of genetic disease diagnostics and treatments: education and equitable implementation.

Visualizing the phenotype diversity: a case study of Alexander disease

Since only a small number of patients have a rare disease, it is difficult to identify all of the features of these diseases. This is especially true for patients uncommonly presenting with rare diseases. It can also be difficult for the patient, their families, and even clinicians to know which one of a number of disease phenotypes the patient is exhibiting. To address this issue, during Biomedical Linked Annotation Hackathon 7 (BLAH7), we tried to extract Alexander disease patient data in Portable Document Format. We then visualized the phenotypic diversity of those Alexander disease patients with uncommon presentations. This led to us identifying several issues that we need to overcome in our future work.

Generalizability of Polygenic Risk Scores for Breast Cancer Among Women With European, African, and Latinx Ancestry

IMPORTANCE

Multiple polygenic risk scores (PRSs) for breast cancer have been developed from large research consortia; however, their generalizability to diverse clinical settings is unknown.

OBJECTIVE

To examine the performance of previously developed breast cancer PRSs in a clinical setting for women of European, African, and Latinx ancestry.

DESIGN, SETTING, AND PARTICIPANTS

This cohort study using the Electronic Medical Records and Genomics (eMERGE) network data set included 39 591 women from 9 contributing medical centers in the US that had electronic medical records (EMR) linked to genotype data. Breast cancer cases and controls were identified through a validated EMR phenotyping algorithm.

MAIN OUTCOMES AND MEASURES

Multivariable logistic regression was used to assess the association between breast cancer risk and 7 previously developed PRSs, adjusting for age, study site, breast cancer family history, and first 3 ancestry informative principal components.

RESULTS

This study included 39 591 women: 33 594 with European, 3801 with African, and 2196 with Latinx ancestry. The mean (SD) age at breast cancer diagnosis was 60.7 (13.0), 58.8 (12.5), and 60.1 (13.0) years for women with European, African, and Latinx ancestry, respectively. PRSs derived from women with European ancestry were associated with breast cancer risk in women with European ancestry (highest odds ratio [OR] per 1-SD increase, 1.46; 95% CI, 1.41-1.51), women with Latinx ancestry (highest OR, 1.31; 95% CI, 1.09-1.58), and women with African ancestry (OR, 1.19; 95% CI, 1.05-1.35). For women with European ancestry, this association with breast cancer risk was largest in the extremes of the PRS distribution, with ORs ranging from 2.19 (95% CI, 1.84-2.53) to 2.48 (95% CI, 1.89-3.25) for the 3 different PRSs examined for those in the highest 1% of the PRS compared with those in the middle quantile. Among women with Latinx and African ancestries at the extremes of the PRS distribution, there were no statistically significant associations.

CONCLUSIONS AND RELEVANCE

This cohort study found that PRS models derived from women with European ancestry for breast cancer risk generalized well for women with European, Latinx, and African ancestries across different clinical settings, although the effect sizes for women with African ancestry were smaller, likely because of differences in risk allele frequencies and linkage disequilibrium patterns. These results highlight the need to improve representation of diverse population groups, particularly women with African ancestry, in genomic research cohorts.

AMELIE speeds Mendelian diagnosis by matching patient phenotype and genotype to primary literature

The diagnosis of Mendelian disorders requires labor-intensive literature research. Trained clinicians can spend hours looking for the right publication(s) supporting a single gene that best explains a patient's disease. AMELIE (Automatic Mendelian Literature Evaluation) greatly accelerates this process. AMELIE parses all 29 million PubMed abstracts and downloads and further parses hundreds of thousands of full-text articles in search of information supporting the causality and associated phenotypes of most published genetic variants. AMELIE then prioritizes patient candidate variants for their likelihood of explaining any patient's given set of phenotypes. Diagnosis of singleton patients (without relatives' exomes) is the most time-consuming scenario, and AMELIE ranked the causative gene at the very top for 66% of 215 diagnosed singleton Mendelian patients from the Deciphering Developmental Disorders project. Evaluating only the top 11 AMELIE-scored genes of 127 (median) candidate genes per patient resulted in a rapid diagnosis in more than 90% of cases. AMELIE-based evaluation of all cases was 3 to 19 times more efficient than hand-curated database-based approaches. We replicated these results on a retrospective cohort of clinical cases from Stanford Children's Health and the Manton Center for Orphan Disease Research. An analysis web portal with our most recent update, programmatic interface, and code is available at AMELIE.stanford.edu.

A pipeline-friendly software tool for genome diagnostics to prioritize genes by matching patient symptoms to literature

Despite an explosive growth of next-generation sequencing data, genome diagnostics only provides a molecular diagnosis to a minority of patients. Software tools that prioritize genes based on patient symptoms using known gene-disease associations may complement variant filtering and interpretation to increase chances of success. However, many of these tools cannot be used in practice because they are embedded within variant prioritization algorithms, or exist as remote services that cannot be relied upon or are unacceptable because of legal/ethical barriers. In addition, many tools are not designed for command-line usage, closed-source, abandoned, or unavailable. We present Variant Interpretation using Biomedical literature Evidence (VIBE), a tool to prioritize disease genes based on Human Phenotype Ontology codes. VIBE is a locally installed executable that ensures operational availability and is built upon DisGeNET-RDF, a comprehensive knowledge platform containing gene-disease associations mostly from literature and variant-disease associations mostly from curated source databases. VIBE's command-line interface and output are designed for easy incorporation into bioinformatic pipelines that annotate and prioritize variants for further clinical interpretation. We evaluate VIBE in a benchmark based on 305 patient cases alongside seven other tools. Our results demonstrate that VIBE offers consistent performance with few cases missed, but we also find high complementarity among all tested tools. VIBE is a powerful, free, open source and locally installable solution for prioritizing genes based on patient symptoms. Project source code, documentation, benchmark and executables are available at https://github.com/molgenis/vibe.

Clinical usefulness of multigene screening with phenotype-driven bioinformatics analysis for the diagnosis of patients with monogenic diabetes or severe insulin resistance

AIMS

Monogenic diabetes is clinically heterogeneous and differs from common forms of diabetes (type 1 and 2). We aimed to investigate the clinical usefulness of a comprehensive genetic testing system, comprised of targeted next-generation sequencing (NGS) with phenotype-driven bioinformatics analysis in patients with monogenic diabetes, which uses patient genotypic and phenotypic data to prioritize potentially causal variants.

METHODS

We performed targeted NGS of 383 genes associated with monogenic diabetes or common forms of diabetes in 13 Japanese patients with suspected (n = 10) or previously diagnosed (n = 3) monogenic diabetes or severe insulin resistance. We performed in silico structural analysis and phenotype-driven bioinformatics analysis of candidate variants from NGS data.

RESULTS

Among the patients suspected having monogenic diabetes or insulin resistance, we diagnosed 3 patients as subtypes of monogenic diabetes due to disease-associated variants of INSR, LMNA, and HNF1B. Additionally, in 3 other patients, we detected rare variants with potential phenotypic effects. Notably, we identified a novel missense variant in TBC1D4 and an MC4R variant, which together may cause a mixed phenotype of severe insulin resistance.

CONCLUSIONS

This comprehensive approach could assist in the early diagnosis of patients with monogenic diabetes and facilitate the provision of tailored therapy.

Mitochondrial Diseases: A Diagnostic Revolution

Mitochondrial disorders have emerged as a common cause of inherited disease, but are traditionally viewed as being difficult to diagnose clinically, and even more difficult to comprehensively characterize at the molecular level. However, new sequencing approaches, particularly whole-genome sequencing (WGS), have dramatically changed the landscape. The combined analysis of nuclear and mitochondrial DNA (mtDNA) allows rapid diagnosis for the vast majority of patients, but new challenges have emerged. We review recent discoveries that will benefit patients and families, and highlight emerging questions that remain to be resolved.

Expansion of the Human Phenotype Ontology (HPO) knowledge base and resources

The Human Phenotype Ontology (HPO)—a standardized vocabulary of phenotypic abnormalities associated with 7000+ diseases—is used by thousands of researchers, clinicians, informaticians and electronic health record systems around the world. Its detailed descriptions of clinical abnormalities and computable disease definitions have made HPO the de facto standard for deep phenotyping in the field of rare disease. The HPO’s interoperability with other ontologies has enabled it to be used to improve diagnostic accuracy by incorporating model organism data. It also plays a key role in the popular Exomiser tool, which identifies potential disease-causing variants from whole-exome or whole-genome sequencing data. Since the HPO was first introduced in 2008, its users have become both more numerous and more diverse. To meet these emerging needs, the project has added new content, language translations, mappings and computational tooling, as well as integrations with external community data. The HPO continues to collaborate with clinical adopters to improve specific areas of the ontology and extend standardized disease descriptions. The newly redesigned HPO website (www.human-phenotype-ontology.org) simplifies browsing terms and exploring clinical features, diseases, and human genes.

Genomic Data Sharing for Novel Mendelian Disease Gene Discovery: The Matchmaker Exchange

In the last decade, exome and/or genome sequencing has become a common test in the diagnosis of individuals with features of a rare Mendelian disorder. Despite its success, this test leaves the majority of tested individuals undiagnosed. This review describes the Matchmaker Exchange (MME), a federated network established to facilitate the solving of undiagnosed rare-disease cases through data sharing. MME supports genomic matchmaking, the act of connecting two or more parties looking for cases with similar phenotypes and variants in the same candidate genes. An application programming interface currently connects six matchmaker nodes-the Database of Chromosomal Imbalance and Phenotype in Humans Using Ensembl Resources (DECIPHER), GeneMatcher, PhenomeCentral, seqr, MyGene2, and the Initiative on Rare and Undiagnosed Diseases (IRUD) Exchange-resulting in a collective data set spanning more than 150,000 cases from more than 11,000 contributors in 88 countries. Here, we describe the successes and challenges of MME, its individual matchmaking nodes, plans for growing the network, and considerations for future directions.

PRIORI-T: A tool for rare disease gene prioritization using MEDLINE

Introduction

Phenotype-driven rare disease gene prioritization relies on high quality curated resources containing disease, gene and phenotype annotations. However, the effectiveness of gene prioritization tools is constrained by the incomplete coverage of rare disease, phenotype and gene annotations in such curated resources.

Methods

We extracted rare disease correlation pairs involving diseases, phenotypes and genes from MEDLINE abstracts and used the information propagation algorithm GCAS to build an association network. We built a tool called PRIORI-T for rare disease gene prioritization that uses this network for phenotype-driven rare disease gene prioritization. The quality of disease-gene associations in PRIORI-T was compared with resources such as DisGeNET and Open Targets in the context of rare diseases. The gene prioritization performance of PRIORI-T was evaluated using phenotype descriptions of 230 real-world rare disease clinical cases collated from recent publications, as well as compared to other gene prioritization tools such as HANRD and Orphamizer.

Results

PRIORI-T contains qualitatively better associations than DisGeNET and Open Targets. Furthermore, the causal genes were captured within Top-50 for more than 40% of the real-world clinical cases and within Top-300 for more than 72% of the cases when PRIORI-T was used for gene prioritization. It outperformed other gene prioritization tools such as HANRD and Orphamizer that primarily rely on curated resources.

Conclusions

PRIORI-T exhibited improved gene prioritization performance without requiring high quality curated data. Thus, it holds great promise in phenotype-driven gene prioritization for rare disease studies.

Diagnosis support systems for rare diseases: a scoping review

INTRODUCTION

Rare diseases affect approximately 350 million people worldwide. Delayed diagnosis is frequent due to lack of knowledge of most clinicians and a small number of expert centers. Consequently, computerized diagnosis support systems have been developed to address these issues, with many relying on rare disease expertise and taking advantage of the increasing volume of generated and accessible health-related data. Our objective is to perform a review of all initiatives aiming to support the diagnosis of rare diseases.

METHODS

A scoping review was conducted based on methods proposed by Arksey and O'Malley. A charting form for relevant study analysis was developed and used to categorize data.

RESULTS

Sixty-eight studies were retained at the end of the charting process. Diagnosis targets varied from 1 rare disease to all rare diseases. Material used for diagnosis support consisted mostly of phenotype concepts, images or fluids. Fifty-seven percent of the studies used expert knowledge. Two-thirds of the studies relied on machine learning algorithms, and one-third used simple similarities. Manual algorithms were encountered as well. Most of the studies presented satisfying performance of evaluation by comparison with references or with external validation. Fourteen studies provided online tools, most of which aimed to support the diagnosis of all rare diseases by considering queries based on phenotype concepts.

CONCLUSION

Numerous solutions relying on different materials and use of various methodologies are emerging with satisfying preliminary results. However, the variability of approaches and evaluation processes complicates the comparison of results. Efforts should be made to adequately validate these tools and guarantee reproducibility and explicability.

Evaluation of Face2Gene using facial images of patients with congenital dysmorphic syndromes recruited in Japan

An increasing number of genetic syndromes present a challenge to clinical geneticists. A deep learning-based diagnosis assistance system, Face2Gene, utilizes the aggregation of "gestalt," comprising data summarizing features of patients' facial images, to suggest candidate syndromes. Because Face2Gene's results may be affected by ethnicity and age at which training facial images were taken, the system performance for patients in Japan is still unclear. Here, we present an evaluation of Face2Gene using the following two patient groups recruited in Japan: Group 1 consisting of 74 patients with 47 congenital dysmorphic syndromes, and Group 2 consisting of 34 patients with Down syndrome. In Group 1, facial recognition failed for 4 of 74 patients, while 13-21 of 70 patients had a diagnosis for which Face2Gene had not been trained. Omitting these 21 patients, for 85.7% (42/49) of the remainder, the correct syndrome was identified within the top 10 suggested list. In Group 2, for the youngest facial images taken for each of the 34 patients, Down syndrome was successfully identified as the highest-ranking condition using images taken from newborns to those aged 25 years. For the oldest facial images taken at ≥20 years in each of 17 applicable patients, Down syndrome was successfully identified as the highest- and second-highest-ranking condition in 82.2% (14/17) and 100% (17/17) of the patients using images taken from 20 to 40 years. These results suggest that Face2Gene in its current format is already useful in suggesting candidate syndromes to clinical geneticists, using patients with congenital dysmorphic syndromes in Japan.