Anat Etzion-Fuchs
Research Subjects
Anat Etzion-Fuchs has a total of 13 co-authors. Here are the top co-authors by number of shared publications:
Adva Wolf
Pheno.AI, Tel-Aviv, Israel
1 shared publication- Alon Diament
Pheno.AI, Tel-Aviv, Israel
1 shared publication - Anna Rice
Pheno.AI, Tel-Aviv, Israel
1 shared publication - Dana Azouri
Pheno.AI, Tel-Aviv, Israel
1 shared publication David A Todd
Department of Computer Science, Princeton University, 35 Olden Street, Princeton, NJ 08544, USA
1 shared publicationDikla Gelbard Solodkin
Pheno.AI, Tel-Aviv, Israel
1 shared publication
Latest Publications
- Open Access
Grounding Health AI: Architecture and Evaluation of a Domain-Expert Metabolic Health Agent
General-purpose language models generate fluent health reports that can fabricate derived clinical met-rics. In an illustrative comparison on identical two-week CGM and meal data, leading foundation models produced reports with invented MAGE values, inflated meal counts, and unreferenced complication-risk projections — failures invisible to non-expert readers and plausible enough to mislead clinicians. We describe the HPP Personal Health Agent (PHA), a metabolic health agent that grounds generation in four layers: the Human Phenotype Project (HPP), a deep-phenotyped cohort of 13,000+ participants sup-plying population references and trained predictive models; 21 domain-expert tools and trained-model wrappers that compute clinical metrics and risk predictions; declarative behavioural skills that constrain what the model may claim; and 21 automated evals across 8 categories developed via a test-driven cycle in which each eval encodes a failure mode discovered during iterative development. In a 210-report ma-trix (14 participants × 3 prompts × 5 system conditions), the gains are largest on the system’s primary use case — meal-grounded metabolic reports, the report it was designed for — where the full system raises a deterministic form/provenance score from 0.37 (the same foundation model with no tools or skills) to 0.91; this score measures structural completeness, numerical accuracy, tool grounding, and clinical-language compliance — a necessary condition for trustworthy health reporting, with clinical quality as a complementary axis examined qualitatively. A skills-vs-tools decomposition shows the two layers act on different axes: tools drive numerical accuracy (≈14% → 90% of reported metrics correct), while the declarative skills add most of the remaining gain in citations, completeness, and structure (tools alone recover only part of the gap, 0.49 from the same 0.37 baseline). The lift generalises beyond the primary use case — to a second metabolic prompt (0.72) and a cardiovascular extension (0.70), each from a 0.37–0.39 baseline. The architecture extends across clinical domains: adding a SCORE2 cardiovascular risk tool and a corresponding skill — with no changes to orchestration, eval harness, or existing tools — produced a cardiovascular risk report from the same system. Trustworthy domain-specialised health AI is a systems design problem: deep-phenotyped cohort data, domain-expert tools and models, and eval-driven development together form a replicable pattern.
Open AccessGrounding Health AI: Architecture and Evaluation of a Domain-Expert Metabolic Health Agent
General-purpose language models generate fluent health reports that can fabricate derived clinical met-rics. In an illustrative comparison on identical two-week CGM and meal data, leading foundation models produced reports with invented MAGE values, inflated meal counts, and unreferenced complication-risk projections — failures invisible to non-expert readers and plausible enough to mislead clinicians. We describe the HPP Personal Health Agent (PHA), a metabolic health agent that grounds generation in four layers: the Human Phenotype Project (HPP), a deep-phenotyped cohort of 13,000+ participants sup-plying population references and trained predictive models; 21 domain-expert tools and trained-model wrappers that compute clinical metrics and risk predictions; declarative behavioural skills that constrain what the model may claim; and 21 automated evals across 8 categories developed via a test-driven cycle in which each eval encodes a failure mode discovered during iterative development. In a 210-report ma-trix (14 participants × 3 prompts × 5 system conditions), the gains are largest on the system’s primary use case — meal-grounded metabolic reports, the report it was designed for — where the full system raises a deterministic form/provenance score from 0.37 (the same foundation model with no tools or skills) to 0.91; this score measures structural completeness, numerical accuracy, tool grounding, and clinical-language compliance — a necessary condition for trustworthy health reporting, with clinical quality as a complementary axis examined qualitatively. A skills-vs-tools decomposition shows the two layers act on different axes: tools drive numerical accuracy (≈14% → 90% of reported metrics correct), while the declarative skills add most of the remaining gain in citations, completeness, and structure (tools alone recover only part of the gap, 0.49 from the same 0.37 baseline). The lift generalises beyond the primary use case — to a second metabolic prompt (0.72) and a cardiovascular extension (0.70), each from a 0.37–0.39 baseline. The architecture extends across clinical domains: adding a SCORE2 cardiovascular risk tool and a corresponding skill — with no changes to orchestration, eval harness, or existing tools — produced a cardiovascular risk report from the same system. Trustworthy domain-specialised health AI is a systems design problem: deep-phenotyped cohort data, domain-expert tools and models, and eval-driven development together form a replicable pattern.
- Open Access
- 17 May 2021
- Published by Oxford University Press (OUP) in Nucleic Acids Research
- Vol. 49 (13) , e78
dSPRINT: predicting DNA, RNA, ion, peptide and small molecule interaction sites within protein domains
Domains are instrumental in facilitating protein interactions with DNA, RNA, small molecules, ions and peptides. Identifying ligand-binding domains within sequences is a critical step in protein function annotation, and the ligand-binding properties of proteins are frequently analyzed based upon whether they contain one of these domains. To date, however, knowledge of whether and how protein domains interact with ligands has been limited to domains that have been observed in co-crystal structures; this leaves approximately two-thirds of human protein domain families uncharacterized with respect to whether and how they bind DNA, RNA, small molecules, ions and peptides. To fill this gap, we introduce dSPRINT, a novel ensemble machine learning method for predicting whether a domain binds DNA, RNA, small molecules, ions or peptides, along with the positions within it that participate in these types of interactions. In stringent cross-validation testing, we demonstrate that dSPRINT has an excellent performance in uncovering ligand-binding positions and domains. We also apply dSPRINT to newly characterize the molecular functions of domains of unknown function. dSPRINT’s predictions can be transferred from domains to sequences, enabling predictions about the ligand-binding properties of 95% of human genes. The dSPRINT framework and its predictions for 6503 human protein domains are freely available at http://protdomain.princeton.edu/dsprint.
- 17 May 2021
- Published by Oxford University Press (OUP) in Nucleic Acids Research
- Vol. 49 (13) , e78
Open AccessdSPRINT: predicting DNA, RNA, ion, peptide and small molecule interaction sites within protein domains
Domains are instrumental in facilitating protein interactions with DNA, RNA, small molecules, ions and peptides. Identifying ligand-binding domains within sequences is a critical step in protein function annotation, and the ligand-binding properties of proteins are frequently analyzed based upon whether they contain one of these domains. To date, however, knowledge of whether and how protein domains interact with ligands has been limited to domains that have been observed in co-crystal structures; this leaves approximately two-thirds of human protein domain families uncharacterized with respect to whether and how they bind DNA, RNA, small molecules, ions and peptides. To fill this gap, we introduce dSPRINT, a novel ensemble machine learning method for predicting whether a domain binds DNA, RNA, small molecules, ions or peptides, along with the positions within it that participate in these types of interactions. In stringent cross-validation testing, we demonstrate that dSPRINT has an excellent performance in uncovering ligand-binding positions and domains. We also apply dSPRINT to newly characterize the molecular functions of domains of unknown function. dSPRINT’s predictions can be transferred from domains to sequences, enabling predictions about the ligand-binding properties of 95% of human genes. The dSPRINT framework and its predictions for 6503 human protein domains are freely available at http://protdomain.princeton.edu/dsprint.