Leonardo Chimirri
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Leonardo Chimirri has a total of 101 co-authors. Here are the top co-authors by number of shared publications:
- Peter Nicholas Robinson
Berlin Institute of Health at Charité Universitätsmedizin , Luisenstr. 65, 10117, Berlin
10 shared publications - Justin Reese
Division of Environmental Genomics and Systems Biology, Lawrence Berkeley National Laboratory, Berkeley, CA, USA
7 shared publications - Melissa A. Haendel
University of North Carolina at Chapel Hill
6 shared publications - Adam S. L. Graefe
Berlin Institute of Health at Charité-Universitätsmedizin Berlin
5 shared publications - Christopher John Mungall
Lawrence Berkeley National Laboratory, Berkeley, USA
5 shared publications - Elena Casiraghi
Department of Computer Science, Aalto University, Espoo, Finland, Department of Computer Science, University of Milan, Milan, Italy, and Environmental Genomics and Systems Biology Division, Lawrence Berkeley National Lab, Berkeley, USA
5 shared publications
Latest Publications
- Open Access
- 2 August 2026
- Published by Elsevier in Neurocomputing
- Vol. 702
Tabular implicit deep neural networks ensembles for the prediction of pathogenic genetic variants in Mendelian diseases
- 2 August 2026
- Published by Elsevier in Neurocomputing
- Vol. 702
Open AccessTabular implicit deep neural networks ensembles for the prediction of pathogenic genetic variants in Mendelian diseases
- Open Access
- 13 March 2026
- Published by Elsevier in SSRN Electronic Journal
Tabular implicit deep neural networks ensembles for the prediction of pathogenic genetic variants in Mendelian diseases
The genetic basis is known for only about half of the approximately 10,000 Mendelian genetic diseases described in the literature, and in most cases, molecular diagnosis remains difficult and elusive. In this context, machine learning methods play a key role. However, identifying pathogenic variants in non-coding regions of the human genome is particularly challenging, as they are vastly outnumbered by neutral variants. This extreme imbalance causes standard machine learning approaches to exhibit strong prediction bias toward the majority (neutral) class, significantly limiting their sensitivity.Building on recent advances in imbalance-aware and ensemble learning methods that have shown superior performance in pathogenic variant prediction, we propose two novel deep learning models to further improve predictive accuracy on this challenging task, a key element for the improvement of the downstream diagnostic yield for patients with rare genetic diseases. The first model, T-ResNet (Tabular Residual Neural Network), adopts a modular architecture with residual connections to support efficient hyperparameter optimization, along with a mini-batch balancing strategy to address class imbalance. The second model, TIDE-Var (Tabular Implicit Deep neural network Ensembles for Variant prediction), is an implicit ensemble of deep neural networks that partially share parameters and in parallel contextually optimize the same objective function.Genome-wide experiments show that T-ResNet achieves an average Area Under the Precision-Recall Curve (AUPRC) comparable to that of HyperSMURF (0.645 ± 0.049), a state-of-the-art method. In contrast, TIDE-Var yields significantly better results (AUPRC = 0.706 ± 0.028). Ablation studies confirm that regularized contextual ensemble learning is a key factor in achieving high predictive performance on this highly imbalanced and complex prediction problem.
- 13 March 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessTabular implicit deep neural networks ensembles for the prediction of pathogenic genetic variants in Mendelian diseases
The genetic basis is known for only about half of the approximately 10,000 Mendelian genetic diseases described in the literature, and in most cases, molecular diagnosis remains difficult and elusive. In this context, machine learning methods play a key role. However, identifying pathogenic variants in non-coding regions of the human genome is particularly challenging, as they are vastly outnumbered by neutral variants. This extreme imbalance causes standard machine learning approaches to exhibit strong prediction bias toward the majority (neutral) class, significantly limiting their sensitivity.Building on recent advances in imbalance-aware and ensemble learning methods that have shown superior performance in pathogenic variant prediction, we propose two novel deep learning models to further improve predictive accuracy on this challenging task, a key element for the improvement of the downstream diagnostic yield for patients with rare genetic diseases. The first model, T-ResNet (Tabular Residual Neural Network), adopts a modular architecture with residual connections to support efficient hyperparameter optimization, along with a mini-batch balancing strategy to address class imbalance. The second model, TIDE-Var (Tabular Implicit Deep neural network Ensembles for Variant prediction), is an implicit ensemble of deep neural networks that partially share parameters and in parallel contextually optimize the same objective function.Genome-wide experiments show that T-ResNet achieves an average Area Under the Precision-Recall Curve (AUPRC) comparable to that of HyperSMURF (0.645 ± 0.049), a state-of-the-art method. In contrast, TIDE-Var yields significantly better results (AUPRC = 0.706 ± 0.028). Ablation studies confirm that regularized contextual ensemble learning is a key factor in achieving high predictive performance on this highly imbalanced and complex prediction problem.
- Top Cited PapersOpen Access
- 24 February 2026
- Published by Springer Nature in European Journal of Human Genetics
- Vol. 34 (4) , 498-504
Systematic benchmarking demonstrates large language models have not reached the diagnostic accuracy of traditional rare-disease decision support tools
Large language models (LLMs) show promise in supporting differential diagnosis, but their performance is challenging to evaluate due to the unstructured nature of their responses, and their accuracy compared to existing diagnostic tools is not well characterized. To assess the current capabilities of LLMs to diagnose genetic diseases, we benchmarked these models on 5213 previously published case reports using the Phenopacket Schema, the Human Phenotype Ontology and Mondo disease ontology. Prompts generated from each phenopacket were sent to seven LLMs, including four generalist models and three LLMs specialized for medical applications. The same phenopackets were used as input to a widely used diagnostic tool, Exomiser, in phenotype-only mode. The best LLM ranked the correct diagnosis first in 23.6% of cases, whereas Exomiser did so in 35.5% of cases. While the performance of LLMs for supporting differential diagnosis has been improving, it has not reached the level of commonly used traditional bioinformatics tools. Future research is needed to determine the best approach to incorporate LLMs into diagnostic pipelines.
- 24 February 2026
- Published by Springer Nature in European Journal of Human Genetics
- Vol. 34 (4) , 498-504
Top Cited PapersOpen AccessSystematic benchmarking demonstrates large language models have not reached the diagnostic accuracy of traditional rare-disease decision support tools
Large language models (LLMs) show promise in supporting differential diagnosis, but their performance is challenging to evaluate due to the unstructured nature of their responses, and their accuracy compared to existing diagnostic tools is not well characterized. To assess the current capabilities of LLMs to diagnose genetic diseases, we benchmarked these models on 5213 previously published case reports using the Phenopacket Schema, the Human Phenotype Ontology and Mondo disease ontology. Prompts generated from each phenopacket were sent to seven LLMs, including four generalist models and three LLMs specialized for medical applications. The same phenopackets were used as input to a widely used diagnostic tool, Exomiser, in phenotype-only mode. The best LLM ranked the correct diagnosis first in 23.6% of cases, whereas Exomiser did so in 35.5% of cases. While the performance of LLMs for supporting differential diagnosis has been improving, it has not reached the level of commonly used traditional bioinformatics tools. Future research is needed to determine the best approach to incorporate LLMs into diagnostic pipelines.
- Open Access
- 23 December 2025
- Published by Elsevier in The American Journal of Human Genetics
- Vol. 113 (1) , 57-70
GA4GH phenopacket-driven characterization of genotype-phenotype correlations in Mendelian disorders
- 23 December 2025
- Published by Elsevier in The American Journal of Human Genetics
- Vol. 113 (1) , 57-70
Open AccessGA4GH phenopacket-driven characterization of genotype-phenotype correlations in Mendelian disorders
- Open Access
- 15 October 2025
- Published by Elsevier in EBioMedicine
- Vol. 121, 105957
Consistent performance of large language models in rare disease diagnosis across ten languages and 4917 cases
- 15 October 2025
- Published by Elsevier in EBioMedicine
- Vol. 121, 105957
Open AccessConsistent performance of large language models in rare disease diagnosis across ten languages and 4917 cases
- Open Access
GA4GH Phenopacket-Driven Characterization of Genotype-Phenotype Correlations in Mendelian Disorders
Comprehensively characterizing genotype-phenotype correlations (GPCs) in Mendelian disease would create new opportunities for improving clinical management and understanding disease biology. However, heterogeneous approaches to data sharing, reuse, and analysis have hindered progress in the field. We developed Genotype Phenotype Evaluation of Statistical Association (GPSEA), a software package that leverages the Global Alliance for Genomics and Health (GA4GH) Phenopacket Schema to represent case-level clinical and genetic data about individuals. GPSEA applies an independent filtering strategy to boost statistical power to detect categorical GPCs represented by Human Phenotype Ontology terms. GPSEA additionally enables visualization and analysis of continuous phenotypes, clinical severity scores, and survival data such as age of onset of disease or clinical manifestations. We applied GPSEA to 85 cohorts with 6613 previously published individuals with variants in one of 80 genes associated with 122 Mendelian diseases and identified 225 significant GPCs, with 48 cohorts having at least one statistically significant GPC. These results highlight the power of standardized representations of clinical data for scalable discovery of GPCs in Mendelian disease.
Open AccessGA4GH Phenopacket-Driven Characterization of Genotype-Phenotype Correlations in Mendelian Disorders
Comprehensively characterizing genotype-phenotype correlations (GPCs) in Mendelian disease would create new opportunities for improving clinical management and understanding disease biology. However, heterogeneous approaches to data sharing, reuse, and analysis have hindered progress in the field. We developed Genotype Phenotype Evaluation of Statistical Association (GPSEA), a software package that leverages the Global Alliance for Genomics and Health (GA4GH) Phenopacket Schema to represent case-level clinical and genetic data about individuals. GPSEA applies an independent filtering strategy to boost statistical power to detect categorical GPCs represented by Human Phenotype Ontology terms. GPSEA additionally enables visualization and analysis of continuous phenotypes, clinical severity scores, and survival data such as age of onset of disease or clinical manifestations. We applied GPSEA to 85 cohorts with 6613 previously published individuals with variants in one of 80 genes associated with 122 Mendelian diseases and identified 225 significant GPCs, with 48 cohorts having at least one statistically significant GPC. These results highlight the power of standardized representations of clinical data for scalable discovery of GPCs in Mendelian disease.
- Open Access
Consistent Performance of GPT-4o in Rare Disease Diagnosis Across Nine Languages and 4967 Cases
Summary: Background: Large language models (LLMs) are increasingly used in the medical field for diverse applications including differential diagnostic support. The estimated training data used to create LLMs such as the Generative Pretrained Transformer (GPT) predominantly consist of English-language texts, but LLMs could be used across the globe to support diagnostics if language barriers could be overcome. Initial pilot studies on the utility of LLMs for differential diagnosis in languages other than English have shown promise, but a large-scale assessment on the relative performance of these models in a variety of European and non-European languages on a comprehensive corpus of challenging rare-disease cases is lacking. Methods: We created 4967 clinical vignettes using structured data captured with Human Phenotype Ontology (HPO) terms with the Global Alliance for Genomics and Health (GA4GH) Phenopacket Schema. These clinical vignettes span a total of 378 distinct genetic diseases with 2618 associated phenotypic features. We used translations of the Human Phenotype Ontology together with language-specific templates to generate prompts in English, Chinese, Czech, Dutch, German, Italian, Japanese, Spanish, and Turkish. We applied GPT-4o, version gpt-4o-2024-08-06, to the task of delivering a ranked differential diagnosis using a zero-shot prompt. An ontology-based approach with the Mondo disease ontology was used to map synonyms and to map disease subtypes to clinical diagnoses in order to automate evaluation of LLM responses. Findings: For English, GPT-4o placed the correct diagnosis at the first rank 19·8% and within the top-3 ranks 27·0% of the time. In comparison, for the eight non-English languages tested here the correct diagnosis was placed at rank 1 between 16·9% and 20·5%, within top-3 between 25·3% and 27·7% of cases. Interpretation: The differential diagnostic performance of GPT-4o across a comprehensive corpus of rare-disease cases was consistent across the nine languages tested. This suggests that LLMs such as GPT-4o may have utility in non-English clinical settings. Funding: NHGRI 5U24HG011449 and 5RM1HG010860. P.N.R. was supported by a Professorship of the Alexander von Humboldt Foundation; P.L. was supported by a National Grant (PMP21/00063 ONTOPREC-ISCIII, Fondos FEDER).
Open AccessConsistent Performance of GPT-4o in Rare Disease Diagnosis Across Nine Languages and 4967 Cases
Summary: Background: Large language models (LLMs) are increasingly used in the medical field for diverse applications including differential diagnostic support. The estimated training data used to create LLMs such as the Generative Pretrained Transformer (GPT) predominantly consist of English-language texts, but LLMs could be used across the globe to support diagnostics if language barriers could be overcome. Initial pilot studies on the utility of LLMs for differential diagnosis in languages other than English have shown promise, but a large-scale assessment on the relative performance of these models in a variety of European and non-European languages on a comprehensive corpus of challenging rare-disease cases is lacking. Methods: We created 4967 clinical vignettes using structured data captured with Human Phenotype Ontology (HPO) terms with the Global Alliance for Genomics and Health (GA4GH) Phenopacket Schema. These clinical vignettes span a total of 378 distinct genetic diseases with 2618 associated phenotypic features. We used translations of the Human Phenotype Ontology together with language-specific templates to generate prompts in English, Chinese, Czech, Dutch, German, Italian, Japanese, Spanish, and Turkish. We applied GPT-4o, version gpt-4o-2024-08-06, to the task of delivering a ranked differential diagnosis using a zero-shot prompt. An ontology-based approach with the Mondo disease ontology was used to map synonyms and to map disease subtypes to clinical diagnoses in order to automate evaluation of LLM responses. Findings: For English, GPT-4o placed the correct diagnosis at the first rank 19·8% and within the top-3 ranks 27·0% of the time. In comparison, for the eight non-English languages tested here the correct diagnosis was placed at rank 1 between 16·9% and 20·5%, within top-3 between 25·3% and 27·7% of cases. Interpretation: The differential diagnostic performance of GPT-4o across a comprehensive corpus of rare-disease cases was consistent across the nine languages tested. This suggests that LLMs such as GPT-4o may have utility in non-English clinical settings. Funding: NHGRI 5U24HG011449 and 5RM1HG010860. P.N.R. was supported by a Professorship of the Alexander von Humboldt Foundation; P.L. was supported by a National Grant (PMP21/00063 ONTOPREC-ISCIII, Fondos FEDER).
- Open Access
Leveraging Generative AI to Accelerate Biocuration of Medical Actions for Rare Disease
Background: Structured representations of clinical data can support computational analysis of individuals and cohorts, and ontologies representing disease entities and phenotypic abnormalities are now commonly used for translational research. The Medical Action Ontology (MAxO) provides a computational representation of treatments and other actions taken for the clinical management of patients. Currently, manual biocuration is used to assign MAxO terms to rare diseases, enabling clinical management of rare diseases to be described computationally for use in clinical decision support and mechanism discovery. However, it is challenging to scale manual curation to comprehensively capture information about medical actions for the more than 10,000 rare diseases. Methods: We present AutoMAxO, a semi-automated workflow that leverages Large Language Models (LLMs) to streamline MAxO biocuration for rare diseases. AutoMAxO first uses LLMs to retrieve candidate curations from abstracts of relevant publications. Next, the candidate curations are matched to ontology terms from MAxO, Human Phenotype Ontology (HPO), and MONDO disease ontology via a combination of LLMs and post-processing techniques. Finally, the matched terms are presented in a structured form to a human curator for approval. Results: We used this approach to process 4,918 unique medical abstracts and identified annotations for 21 rare genetic diseases, we extracted 18,631 candidate disease-treatment curations, 538 of which were confirmed and transferred to the MAxO annotation dataset. Conclusion: The results of this project underscore the potential of generative AI to accelerate precision medicine by enabling a robust and comprehensive curation of the primary literature to represent information about diseases and procedures in a structured fashion. Although we focused on MAxO in this project, similar approaches could be taken for other biomedical curation tasks.
Open AccessLeveraging Generative AI to Accelerate Biocuration of Medical Actions for Rare Disease
Background: Structured representations of clinical data can support computational analysis of individuals and cohorts, and ontologies representing disease entities and phenotypic abnormalities are now commonly used for translational research. The Medical Action Ontology (MAxO) provides a computational representation of treatments and other actions taken for the clinical management of patients. Currently, manual biocuration is used to assign MAxO terms to rare diseases, enabling clinical management of rare diseases to be described computationally for use in clinical decision support and mechanism discovery. However, it is challenging to scale manual curation to comprehensively capture information about medical actions for the more than 10,000 rare diseases. Methods: We present AutoMAxO, a semi-automated workflow that leverages Large Language Models (LLMs) to streamline MAxO biocuration for rare diseases. AutoMAxO first uses LLMs to retrieve candidate curations from abstracts of relevant publications. Next, the candidate curations are matched to ontology terms from MAxO, Human Phenotype Ontology (HPO), and MONDO disease ontology via a combination of LLMs and post-processing techniques. Finally, the matched terms are presented in a structured form to a human curator for approval. Results: We used this approach to process 4,918 unique medical abstracts and identified annotations for 21 rare genetic diseases, we extracted 18,631 candidate disease-treatment curations, 538 of which were confirmed and transferred to the MAxO annotation dataset. Conclusion: The results of this project underscore the potential of generative AI to accelerate precision medicine by enabling a robust and comprehensive curation of the primary literature to represent information about diseases and procedures in a structured fashion. Although we focused on MAxO in this project, similar approaches could be taken for other biomedical curation tasks.
- Open Access
Systematic benchmarking demonstrates large language models have not reached the diagnostic accuracy of traditional rare-disease decision support tools
Large language models (LLMs) show promise in supporting differential diagnosis, but their performance is challenging to evaluate due to the unstructured nature of their responses and their accuracy compared to existing diagnostic tools is not well characterized. To assess the current capabilities of LLMs to diagnose genetic diseases, we benchmarked these models on 5,213 case reports using the Phenopacket Schema, the Human Phenotype Ontology and Mondo disease ontology. Prompts generated from each phenopacket were sent to seven LLMs, including four generalist models and three LLMs specialized for medical applications. The same phenopackets were used as input to a widely used diagnostic tool, Exomiser, in phenotype-only mode. The best LLM ranked the correct diagnosis first in 23.6% of cases, whereas Exomiser did so in 35.5% of cases. While the performance of LLMs for supporting differential diagnosis has been improving, it has not reached the level of commonly used traditional bioinformatics tools. Future research is needed to determine the best approach to incorporate LLMs into diagnostic pipelines.
Open AccessSystematic benchmarking demonstrates large language models have not reached the diagnostic accuracy of traditional rare-disease decision support tools
Large language models (LLMs) show promise in supporting differential diagnosis, but their performance is challenging to evaluate due to the unstructured nature of their responses and their accuracy compared to existing diagnostic tools is not well characterized. To assess the current capabilities of LLMs to diagnose genetic diseases, we benchmarked these models on 5,213 case reports using the Phenopacket Schema, the Human Phenotype Ontology and Mondo disease ontology. Prompts generated from each phenopacket were sent to seven LLMs, including four generalist models and three LLMs specialized for medical applications. The same phenopackets were used as input to a widely used diagnostic tool, Exomiser, in phenotype-only mode. The best LLM ranked the correct diagnosis first in 23.6% of cases, whereas Exomiser did so in 35.5% of cases. While the performance of LLMs for supporting differential diagnosis has been improving, it has not reached the level of commonly used traditional bioinformatics tools. Future research is needed to determine the best approach to incorporate LLMs into diagnostic pipelines.