Mor Peleg

Background: Adverse events (AEs) related to cancer treatment represent a valuable source of information that can be used to adjust therapy for individual patients. The National Institutes of Health developed the Common Terminology Criteria for Adverse Events (CTCAE), a comprehensive standardized terminology for health care providers to consistently report AEs during patient visits. Mobile health technologies, in principle, also allow AEs to be self-reported by patients in between visits; however, the terminology poses challenges for them, both in selecting the correct symptom to report and in rating its severity. The National Institutes of Health developed the Patient-Reported Outcomes–CTACE as the patient-oriented companion of the CTCAE. However, it shows some weaknesses in completeness and precision when used for continuous home patient monitoring and for decision support. Objective: The aim of this paper is to propose a new terminology for reporting AEs that is easy for patients to use while also being clinically meaningful for health care providers and easily exploitable by decision support systems. Moreover, we aim to demonstrate its implementation and validation within the CAPABLE (Cancer Patients Better Life Experience) EU project. Methods: The development of the new terminology starts from the CTCAE, which includes a comprehensive list of signs and symptoms along with guidance for accurately grading their severity. Through a multistep, participatory approach involving both patients and health care providers, we reduced and adapted the AE list for patient-oriented apps. During the CAPABLE project, the proposed terminology was integrated into a mobile app and evaluated within a clinical pilot study involving 86 patients who were monitored through the app for at least 6 months, and a control cohort of 133 patients monitored using standard care practices. Results: The final terminology includes 124 AEs, 49 expressed as “present or absent,” and 77 associated with 4 description levels. A mapping between the description levels and the original CTCAE grades enables running the decision support system embedded in the CAPABLE app. The pilot study demonstrated that the majority of the patients used the symptoms reporting functionality, sharing also 24 unique AEs that are not present in the Patient-Reported Outcomes–CTCAE. Symptoms reported using the proposed terminology allowed the enactment of the clinical practice guidelines included in the CAPABLE decision support tool, triggering 11 distinct recommendations. Conclusions: The results obtained from the clinical study support our claim regarding the need for a novel terminology for the self-reporting of AEs, characterized by ease of use, completeness, and clinical meaningfulness. Finally, by mapping our terminology to the CTCAE, we demonstrated that it is possible to exploit self-reported data to trigger decision support rules consistent with clinical practice guidelines.

Therapy recommendation for chronic patients with multimorbidity is challenging due to risks of treatment conflicts. Existing decision support systems face scalability limitations. Inspired by the way in which general practitioners (GP) manage multimorbidity patients, occasionally convening multidisciplinary team (MDT) collaboration, this study investigated the feasibility and value of using a Large Language Model (LLM)-based multi-agent system (MAS) for safer therapy recommendations. We designed a single agent and a MAS framework simulating MDT decision-making by enabling discussion among LLM agents to resolve medical conflicts. The systems were evaluated on therapy planning tasks for multimorbidity patients using benchmark cases. We compared MAS performance with single-agent approaches and real-world benchmarks. An important contribution of our study is the definition of evaluation metrics that go beyond the technical precision and recall and allow the inspection of clinical goals met and medication burden of the proposed advices to a gold standard benchmark. Our results show that with current LLMs, a single agent GP performs as well as MDTs. The best-scoring models provide correct recommendations that address all clinical goals, yet the advices are incomplete. Some models also present unnecessary medications, resulting in unnecessary conflicts between medication and conditions or drug-drug interactions.

Brain age gap (BAG), the deviation between estimated brain age and chronological age, is a promising marker of brain health. However, the genetic architecture and reliable targets for brain aging remains poorly understood. In this study, we estimate magnetic resonance imaging (MRI)–based brain age using deep learning models trained on the UK Biobank and validated with three external datasets. A genome-wide association study for BAG identified two unreported loci and seven previously reported loci. By integrating Mendelian Randomization (MR) and colocalization analysis on eQTL and pQTL data, we prioritized seven genetically supported druggable genes, including MAPT , TNFSF12 , GZMB , SIRPB1 , GNLY , NMB , and C1RL , as promising targets for brain aging. We rediscovered 13 potential drugs with evidence from clinical trials of aging and prioritized several drugs with strong genetic support. Our study provides insights into the genetic basis of brain aging, potentially facilitating drug development for brain aging to extend the health span.

Background: Patients with melanoma receiving immunotherapy with immune-checkpoint inhibitors often experience immune-related adverse events, cancer-related fatigue, and emotional distress, affecting health-related quality of life () and clinical outcome to immunotherapy. eHealth tools can aid patients with cancer in addressing issues, such as adverse events and psychosocial well-being, from various perspectives. Objective: This study aimed to explore the effect of the Cancer Patients Better Life Experience (CAPABLE) system, accessed through a mobile app, on HRQoL compared with a matched historical control group receiving standard care. CAPABLE is an extensively tested eHealth app, including educational material, remote symptom monitoring, and well-being interventions. Methods: This prospective pilot study compared an exploratory cohort that received the CAPABLE smartphone app and a multisensory smartwatch for 6 months (intervention) to a 2:1 individually matched historical prospective control group. HRQoL data were measured with the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-Core 30 at baseline (T0), 3 months (T1), and 6 months (T2) after start of treatment. Mixed effects linear regression models were used to compare HRQoL between the 2 groups over time. Results: From the 59 eligible patients for the CAPABLE intervention, 31 (53%) signed informed consent to participate. Baseline HRQoL was on average 10 points higher in the intervention group compared with controls, although equally matched on baseline and clinical characteristics. When correcting for sex, age, disease stage, and baseline scores, an adjusted difference in fatigue of −5.09 (95% CI −15.20 to 5.02, P=.32) at month 3 was found. No significant nor clinically relevant adjusted differences on other HRQoL domains over time were found. However, information satisfaction was significantly higher in the CAPABLE group (β=8.71, 95% CI 1.54‐15.88, P=.02). Conclusions: The intervention showed a limited effect on HRQoL, although there was a small improvement in fatigue at 3 months, as well as information satisfaction. When aiming at personalized patient and survivorship care, further optimization and prospective investigation of eHealth tools is warranted.

Ontologies play an important role in the organization and representation of knowledge. However, in most cases, ontologies do not fully cover domain knowledge, resulting in a gap. This gap, often expressed as a lack of concepts, relations, or axioms, is usually filled by domain experts in a manual and tedious process. Utilizing large language models (LLMs) can ease this process; a fine-tuned LLM could receive as input up-to-date and reliable domain knowledge natural text and output a structured graph in OWL RDF/Turtle format, which is the standard format of ontologies. Thus, to fine-tune a model, text-owl sentence pairs that constitute such a dataset must be acquired. Unfortunately, such a dataset does not exist in the literature or within the open-source community. Therefore, this paper introduces our LLM-assisted verbalizer to create such a data set by converting OWL statements from existing ontologies into natural text. We evaluate the verbalizer on 322 classes from four different ontologies using two different LLMs, achieving precision and recall as high as 0.99 and 0.96, respectively.