Rare diseases represent one of the most challenging settings for clinical decision-making, where heterogeneous presentations, sparse evidence and limited expertise create persistent uncertainty throughout the care pathway. Although artificial intelligence could help, existing systems largely address isolated tasks, particularly diagnosis, and usually rely on downstream investigations rather than information available at initial presentation. Here we show that clinical AI performance under uncertainty can be improved not by scaling a single model, but by exploiting the diversity of multiple imperfect reasoning systems. Across heterogeneous large language models, we identify divergent reasoning trajectories with complementary error patterns and develop RareLens, which learns to reconcile these perspectives into actionable decisions across four stages of rare disease care: risk screening, diagnosis, treatment planning and prognosis prediction. Built on RarelensBench, a real-world dataset of 157,525 cases spanning all 33 Orphanet categories and more than 7,000 conditions, RareLens outperformed every frontier model tested, including GPT-5, DeepSeek-R1, Claude-3.7-Sonnet and Gemini-2.5-Pro, across all stages. It achieved an area under the curve of 0.917 for screening and top-1 accuracies of 65.5% and 89.8% for diagnosis and treatment. In an external evaluation involving 1,287 cases and 23 physicians, autonomous RareLens and physicians assisted by RareLens both outperformed unaided physicians, while demonstrating that effective human-AI collaboration requires more than simply providing model outputs. These findings establish divergent model reasoning as an exploitable source of information and suggest a general strategy for building AI systems that operate reliably under high clinical uncertainty.
Uncommon and off-guideline cases are difficult for clinical decision support, because physicians must make a series of management decisions under diagnostic uncertainty and rarely see the full case at once. Most large language model (LLM) benchmarks for medicine score only the final diagnosis, yet much of clinical care turns on the next appropriate action: the next test to order, the imaging study to obtain, the specialist to involve, or the differential to pursue. We introduce MedUPSQA, a dataset of 21,874 mid-stream clinical decision points built from 5,535 real case reports, and MedUPS, an alignment framework that supervises models on these intermediate decisions as they unfold along a patient's trajectory. We segment free-text case presentations into chronologically ordered, accumulating clinical chunks and align models to predict the next step with reinforcement learning (GRPO), using an external LLM-as-a-Judge reward. This objective mirrors how clinicians actually meet patients, reasoning forward from accumulating evidence toward the next decision, rather than committing to a final label. Across three backbones, mid-stream alignment raises next-step accuracy from 55.2 to 66.7 for Qwen3.6-27B, from 47.2 to 57.8 for Qwen3.5-9B, and from 37.8 to 44.4 for HuatuoGPT-3-8B, with 95% CI. In several model scales we test the objective improves accuracy more than scale, with smaller models surpassing larger, frontier models we evaluate. We further train supervised fine-tuning (SFT) baselines on the mid-stream task, SFT improves all backbones above base, indicating the target framwork carries signal independently of the optimizer. We release the dataset, code, and aligned checkpoints.
Ofir Ben Shoham, O. Perets, Nir Grinberg et al.· 0 citations
Genome sequencing enables accurate detection of genetic variants and is transforming rare disease diagnostics. While data generation is scalable, prioritization and clinical interpretation remain challenging, often requiring expert manual classification. AI-driven decision support systems are therefore needed to assist in causal variant identification or to fully automate large-scale re-analysis of unsolved cases. Existing tools often estimate variant impact on protein function, but few integrate genomic, phenotypic, and clinical annotation data for diagnosis. We present aiDIVA, an ensemble-AI combining statistical and machine learning models trained on genomic and phenotypic data to identify causal variants among tens of thousands per patient. aiDIVA applies a random forest model to classify pathogenicity and generates evidence-based scores for dominant and recessive diseases. These predictions are integrated with clinical metadata to prioritize the most likely causal variants. Large language models further refine and explain results. The aiDIVA-meta model consolidates all scores into a ranked list. aiDIVA-meta reported the causal variant among the top-3 candidates in 97.4% of a pre-training collected cohort with prior evidence in ClinVar or HGMD, and in 93.3% of a post-training collected cohort of previously unreported variants.
D. Boceck, L. Laugwitz, Marc Sturm et al.· npj Genomic Medicine· 0 citations
Motivation: Rare disease (RD) diagnosis is frequently delayed due to the similarities in symptoms to common disease variants. Machine Learning Algorithms applied to Electronic Health Records show promise for accelerating the diagnosis; however, legal and privacy concerns pose significant barriers. To address these issues, Synthetic Data Generation is an alternative method for obtaining Electronic Health Records and can be applied with any Machine Learning algorithm for benchmarking and development purposes. Despite the availability of Synthetic Data Generation algorithms, support for generating a subset of patients that differ in a definable degree from the majority to simulate patients with RD is often lacking. Results: We present SYNRARE, a graphical user interface based on the Synthea framework that enables easier modification and generation of synthetic Electronic Health Records of RD patients, which differ only to a definable degree from patients with common diseases, thereby enabling the benchmarking and testing of algorithms under controlled technical conditions. SYNRARE enables researchers to rapidly benchmark their Machine Learning algorithms across any scenario. Availability and implementation: SYNRARE, including detailed instructions for installing, is available at https://gitlab.sdu.dk/screen4care/synrare.
Nicolai Dinh Khang Truong, Richard Rottger· 0 citations
Due to the limited amount of information, modeling longitudinal rare-disease data can benefit from integrating clinical knowledge. Yet, elicitation of expert knowledge and formalization for model fitting is challenging, in particular due to limited time of clinical experts. To nevertheless make domain knowledge accessible during model fitting, we use large language models (LLMs) as synthetic clinical experts to supervise a variational-autoencoder-based approach that learns low-dimensional latent summaries of visit-level observations. Specifically, LLMs are queried offline on textual descriptions of patient observations to obtain judgments, e.g., the suspected clinical category. To improve the variational autoencoder fit, we train a differentiable surrogate model on these judgments and augment the loss function to encourage reconstructions that preserve the clinical-label distribution of their corresponding input profile. In an application to longitudinal motor-function assessments from children with spinal muscular atrophy, we map visit-level clinical profiles to low-dimensional representations that are linked by a multivariate mixed-effects model. The synthetic expert loss discourages reconstructions that remain numerically close in data space but alter the clinical interpretation of the reconstructed motor function profile, such as by crossing a disease-type boundary. We thus reduced disagreement between original and reconstructed SMA type labels from about 11 to 7 percent. Furthermore, informing the latent representation by the synthetic expert improved prediction of motor function milestones compared with unsupervised latent representations and a data-level baseline. These results suggest that incorporating LLMs into model fitting can make clinical knowledge available to representation learning and improve clinical faithfulness for longitudinal rare-disease data.
Clemens Schächter, A. Pechmann, Janbernd Kirschner et al.· 0 citations
BACKGROUND
Prior biological knowledge and phenotype information can help identify disease genes from whole genome/exome sequencing studies, but how best to incorporate external knowledge with variant data remains challenging. We developed a machine learning algorithm called RankVar to prioritize causative variants for rare diseases, based on clinical notes and genome/exome sequencing profiles.
METHODS
RankVar uses a random forest classifier trained on ~ 1 million variants from the 1000 Genomes Project with spiked-in pathogenic variants. For testing, we compiled sequencing data and phenotype information from several independent datasets: 260 subjects from the Children's Hospital of Philadelphia (CHOP) with positive genetic diagnosis of various Mendelian diseases, 135 subjects from Birth Defects Biorepository (BDB), as well as 356 and 97 subjects with candidate causal variants for autism spectrum disorders from the Simons Simplex Collection (SSC) and the Simons Foundation Powering Autism Research for Knowledge (SPARK), respectively.
RESULTS
RankVar achieves a top 10 variant accuracy of 90.0%, 81.5%, 46.1%, and 76.3% for CHOP, BDB, SSC, and SPARK, respectively, with improved performance over existing approaches. Notably, RankVar successfully identified X-linked and Y-linked disease-causal variants, such as KDM6A (p.N915Kfs5*) and SRY (p.W98X), as the top candidate variants. Moreover, we evaluated RankVar for genomic reinterpretation of 130 unsolved CHOP cases with hearing loss and successfully identified 61 candidate causal variants after manual review.
CONCLUSIONS
In summary, RankVar performed favorably relative to existing methods in our evaluation, accommodated different genetic models and X/Y chromosome variants, and may provide a useful framework for prioritizing variants in monogenic or oligogenic diseases. We anticipate that RankVar may aid in primary genetic diagnosis, genome reinterpretation of previously unsolved cases, and the discovery of novel disease genes.
Yuan Zhang, Mian Umair Ahsan, Peng Wang et al.· Genome Medicine· 1 citation
Large language models achieve high scores on medical knowledge assessments, yet clinical reasoning requires actively deciding what to investigate under uncertainty. We developed an agentic evaluation framework in hematologic oncology in which models must proactively request clinical data across three sequential rounds before committing to a diagnosis and treatment plan. Across 32 frontier models, the best achieved only 68% overall accuracy. Information utilization, the fraction of available data actually requested, was the strongest predictor of diagnostic accuracy (R = 0.69, P<0.001), yet utilization collapsed from 57% to 26% in the final round, leaving molecular and cytogenetic data critical for treatment selection unexamined. Reasoning traces scored high on a clinical reasoning rubric (91% above threshold) but decorrelated from accuracy, revealing a gap between locally coherent rationales and globally correct conclusions. Error analysis identified search satisficing, anchoring and premature closure as the dominant failure modes, the same cognitive biases that characterize novice clinicians under dual-process models of diagnostic reasoning. These findings demonstrate that the primary limitation of current models in clinical oncology is not insufficient medical knowledge but a systematic failure of information-seeking under uncertainty.
K. Braitsch, L. Schmalbrock, Theresa Weltermann et al.· 1 citation