OBJECTIVES: Artificial Intelligence models are increasingly used in health care, yet global performance metrics can mask variations in reliability across individual patients or subgroups with shared attributes, called patient profiles. This study introduces predictive performance precision analysis in medicine (MED3pa), a method that identifies when models are less reliable, allowing clinicians to better assess model limitations. MATERIALS AND METHODS: We propose a framework that estimates predictive confidence using 3 combined approaches: individualized (IPC), aggregated (APC), and mixed predictive confidence (MPC). Individualized predictive confidence estimates confidence for each patient, APC assesses it across profiles, and MPC combines both. We evaluate our method on 4 datasets: 1 simulated, 2 public, and 1 private clinical dataset. Metrics by declaration rate curves show how performance changes when retaining only the most confident predictions, while interpretable decision trees reveal profiles with higher or lower model confidence. RESULTS: We demonstrate our method in internal, temporal, and external validation settings, as well as through a clinical example. In internal validation, limiting predictions to the 93% most confident cases improved sensitivity by 14.3% and the area under the receiver operating characteristic curve by 5.1%. In the clinical example, MED3pa identified a patient profile with high misclassification risk, demonstrating its potential for safer deployment. DISCUSSION: By identifying low-confidence predictions, our framework improves model reliability in clinical settings. It can be integrated into decision support systems to help clinicians make more informed decisions. Confidence thresholds help balance model performance with the proportion of patients for whom predictions are considered reliable. CONCLUSION: Better leveraging confidence in model predictions could improve reliability and trustworthiness, supporting safer and more effective use in health care.

We explore the concept of meaning in applied ontologies using possible world semantics. We begin by analyzing the nature of possible worlds in the influential framework proposed by Guarino, Oberle and Staab: interpreting an ontology involves selecting, from the set of logically possible worlds, a subset of worlds that are metaphysically possible according to the ontology – ideally corresponding to the set of worlds that are metaphysically possible according to the ontology’s creator. We argue that this framework is limited to analytic statements and should be extended to encompass synthetic statements. The analytic/synthetic distinction, we suggest, can itself be understood in terms of the necessary/contingent distinction using metaphysically possible worlds. We propose a dual framework for introducing terms in ontology development, integrating both descriptivism and Kripke’s theory of rigid designators. This framework accommodates a posteriori analytic statements, implying that the meaning of a term may be unknown even to its creator. Finally, we distinguish two distinct roles that labels can play in ontology development: either as rigid designators that carry semantic weight, or as mere human-readable tags serving as proxies for underlying descriptions.

According to meaning holism, the meanings of all the words in a language are interdependent. If this was true, then the very practice of building largely interconnected set of ontologies would be threatened. We examine here the extent of the severity of meaning holism for ontology engineering, based on a definition of the meaning of a class term in an ontology, with regard to the classical analytic/synthetic distinction. We show that meaning holism is not as pervasive in ontologies as traditionally assumed in philosophy of language when interpreting the meaning of a class term as a collection of statements expressing necessary conditions on this term. Still, meaning holism presents substantial challenges for ontology engineering and requires mitigation strategies. We also investigate the related phenomenon of indeterminacy of reference and show how anchoring formal ontologies in natural language can mitigate this problem, even if not fully control it.

According to meaning holism, the meanings of all of the words in a language are interdependent. If this was true, then the very practice of building largely interconnected set of ontologies would be threatened. We examine here the extent of the severity of meaning holism for ontology engineering, based on several definitions of the meaning of a class term in an ontology, with regard to the classical analytic/synthetic distinction. We show that meaning holism is not as pervasive in ontologies as traditionally assumed in philosophy of language, and that a conception of meaning of a class term as a collection of statements expressing necessary conditions on this term limits meaning holism further. Still, meaning holism presents substantial challenges for ontology engineering and requires mitigation strategies. We also investigate the related phenomenon of indeterminacy of reference and show how anchoring formal ontologies in natural language can mitigate this problem, even if not fully control it.

Learning Healthcare Systems are an emerging approach to healthcare research as translated into practice. For this purpose, a strong interconnection comes to be a necessity when dealing with healthcare services, research and knowledge transfer all at once. Practically, these connections imply that a routing protocol should guarantee anonymity to entities in compliance with both laws and ethical requirements while restricting the quantity of information obtainable had an entity been compromised. In order to bring more protection and meet all the requirements, a new message routing protocol is offered to allow the use of data access paths and to resist traffic analysis security threats. The protocol protects the addresses and roles pertaining to entities from any lurking malevolent minds by implementing proxies into a mix-network. Moreover, flows of synthetic datasets and contents identifiers are handled separately so as to curb any risk of re-identification. A model of this protocol is provided in the form of a multi-objective optimization problem, natively integrating objectives of minimizing both latency and entropy of the information exchanged. The assessment of this model shows that the constrained separation of data flows has a minimal impact on delay times, which not only reveals to be an acceptable compromise but also significantly increases security in data access.

Realizable entities are properties that can be realized in processes of specific correlated types in which the bearer participates. It will be valuable to create a systematic classification of realizable entities because they are useful for various modeling purposes in ontologies. In this paper we outline a unifying framework for realizable entities (including dispositions and roles) in the upper ontology Basic Formal Ontology (BFO) that is theoretically underpinned by J. McKitrick’s pragmatic approach to dispositions. In particular, we develop a formal ontological account of “extrinsic dispositions” and illustrate its potential applications with clarification of functions and roles in BFO.

OBJECTIVE: The study sought to evaluate the expected clinical utility of automatable prediction models for increasing goals-of-care discussions (GOCDs) among hospitalized patients at the end of life (EOL). MATERIALS AND METHODS: We built a decision model from the perspective of clinicians who aim to increase GOCDs at the EOL using an automated alert system. The alternative strategies were 4 prediction models-3 random forest models and the Modified Hospital One-year Mortality Risk model-to generate alerts for patients at a high risk of 1-year mortality. They were trained on admissions from 2011 to 2016 (70 788 patients) and tested with admissions from 2017-2018 (16 490 patients). GOCDs occurring in usual care were measured with code status orders. We calculated the expected risk difference (beneficial outcomes with alerts minus beneficial outcomes without alerts among those at the EOL), the number needed to benefit (number of alerts needed to increase benefit over usual care by 1 outcome), and the net benefit (benefit minus cost) of each strategy. RESULTS: Models had a C-statistic between 0.79 and 0.86. A code status order occurred during 2599 of 3773 (69%) hospitalizations at the EOL. At a risk threshold corresponding to an alert prevalence of 10%, the expected risk difference ranged from 5.4% to 10.7% and the number needed to benefit ranged from 5.4 to 10.9 alerts. Using revealed preferences, only 2 models improved net benefit over usual care. A random forest model with diagnostic predictors had the highest expected value, including in sensitivity analyses. DISCUSSION: Prediction models with acceptable predictive validity differed meaningfully in their ability to improve over usual decision making. CONCLUSIONS: An evaluation of clinical utility, such as by using decision curve analysis, is recommended after validating a prediction model because metrics of model predictiveness, such as the C-statistic, are not informative of clinical value.

This article provides the basis of a formal axiomatic system for a mereology of informational entities based on the idea of information fillers that can occupy information slots, such as the same word that can be used in different sentences. It is inspired by Karen Bennett’s mereological system that enables a whole to have a part “twice over”, but differs from it in several key points, such as the acceptance of empty slots, and the possibility for slots to have slots. Information slots are analyzed as informational entities that can carry aboutness.

Learning Healthcare System (LHS) is an increasingly deployed approach in health to improve patient care. For the successful implementation of this approach, communications must become cross-cutting between research and primary care. To meet this need, standardized protocols for health data exchange, such as Fast Healthcare Interoperability Resources from Health Level Seven organization, are massively used in healthcare organizations. However, these protocols don’t meet new security needs and they don’t natively integrate anonymization mechanisms for data sources and patients while maintaining individuation. In this paper, a new protocol suite is proposed for sensitive health data exchange. Thus, an architecture is presented: it integrates proxies and anonymizers for the extraction and transmission phases of sensitive data. Then, requirements on several new protocols are detailed to meet the exchanges needs between the learning health system entities. Finally, a comparison of security properties and a vulnerability analysis are carried out between the Fast Healthcare Interoperability Resources protocol and the protocol suite proposed. These analyses show that the protocol suite integrates most of the defenses against common protocol attacks and that anonymization, confidentiality, authentication and logging requirements are met.

Clear criteria for the identity of dispositions are still lacking, and this has been presented as one of the main challenge raised by such entities. It is of prime importance to identify or distinguish dispositions such as diseases or risks. This article first introduces conventional ways to refer to a disposition (such as “fragility”) and canonical ways (such as “disposition to break in case of a strong shock”). This raises the issue of how should exactly be defined a “disposition d to R when TR”, where R is a realization specification and TR a trigger specification. Two ontological frameworks are distinguished. The first framework, which has been largely used so far in the literature on dispositions, interprets d as a disposition which can only be triggered by instances of TR, and can only be realized by instances of R. The second, new framework introduces the notion of “minimal trigger” and “maximal realization”, and interprets TR as a parent class of a class of processes that have as part a minimal trigger, and R as a parent class of a class of processes that are parts of a maximal realization. We then discuss several criteria of identity, including the criterion according to which two dispositions are identical iff they have the same categorical basis, the same class of minimal triggers and the same class of maximal realizations. We show on several examples that the second framework avoids the disposition multiplicativism that is introduced by the first framework.

This article presents a formalization of velocity in the context of a realist and perspectivalist upper ontology like BFO. It argues that the term “velocity” can refer to two different entities: a motion-velocity, which is a process profile characterizing a motion process; and an object-velocity, which is a disposition inhering in the moving object. Three different kinds of motion-velocity are presented: left-velocity, right-velocity and bilateral velocity. Motion-velocity could exist without object-velocity, as revealed by a thought experiment presented by Tooley; but in our world, Newton's first law of inertia implies that every object has both an inertial disposition and a closely related but different disposition that we call “object-velocity.” Those two dispositions are realized by the right-velocity. The left-velocity is a trigger of the inertial disposition, and brings into existence the object-velocity.

This article presents CVDO, an ontology of cardiovascular diseases structured on OBO foundry's principle and based on BFO and FMA. CVDO reorganizes and completes DOID cardiovascular diseases around OGMS tripartite model of disease, and builds its taxonomy of diseases largely by automatic reasoning. It points to the need of OGMS to be supplemented by methodological rules to determine the end point of a disease course, and to locate the material basis of a disease in a causal chain of disorders.

OBJECTIVE: Biomedical research increasingly relies on the integration of information from multiple heterogeneous data sources. Despite the fact that structural and terminological aspects of interoperability are interdependent and rely on a common set of requirements, current efforts typically address them in isolation. We propose a unified ontology-based knowledge framework to facilitate interoperability between heterogeneous sources, and investigate if using the LexEVS terminology server is a viable implementation method. MATERIALS AND METHODS: We developed a framework based on an ontology, the general information model (GIM), to unify structural models and terminologies, together with relevant mapping sets. This allowed a uniform access to these resources within LexEVS to facilitate interoperability by various components and data sources from implementing architectures. RESULTS: Our unified framework has been tested in the context of the EU Framework Program 7 TRANSFoRm project, where it was used to achieve data integration in a retrospective diabetes cohort study. The GIM was successfully instantiated in TRANSFoRm as the clinical data integration model, and necessary mappings were created to support effective information retrieval for software tools in the project. CONCLUSIONS: We present a novel, unifying approach to address interoperability challenges in heterogeneous data sources, by representing structural and semantic models in one framework. Systems using this architecture can rely solely on the GIM that abstracts over both the structure and coding. Information models, terminologies and mappings are all stored in LexEVS and can be accessed in a uniform manner (implementing the HL7 CTS2 service functional model). The system is flexible and should reduce the effort needed from data sources personnel for implementing and managing the integration.