LOGIC-BASED MEDICINE VERSUS EVIDENCE-BASED MEDICINE FOR MODELING QUALIFIED-SELF HEALTH KITS
PATRIK EKLUND
INTRODUCTION
In this chapter we propose to use formal logic in order to bridge the gap between information management in Qualified-Self apps and information classification and structures residing within health and healthcare ontology. Lative logic40 embraces signatures, terms and sentences arising from monads and functors in category theory, and can be arranged in order to enable well- founded logical and ontology representation. modeling uses these theoretical notions in order to extend the logical structure of classifications of health.
Our focus is on Active and Healthy Ageing (AHA) including aspects of assessment (Eklund, 2009) and classification.
Qualified-self aspects within AHA requires having emphasis on empowerment and how citizens as individuals and patients can manage their own data, in particular for self-monitoring purposes. For this management to meet reply properly to the societal grand challenge of AHA, there is the need to shift from society owning all individual health data to individuals themselves owning their data. Another aspect is that the Quantified-Self movement is still rooted mostly in wellness and even fitness, and as having various apps at their disposal. Focus is then not always just on health but on performance more in general.
Apple’s Health Kit was launched to promote such self-qualification, and is made at least for the purpose of further promoting the use of Apple’s mobile phone. Various technology partners have been foreseen to become included, in particular as far as medical record systems and other related registries are concerned. The GoogleFit and Samsung’s Sami follow similar ambitions and potential bindings. Google’s approach is more device oriented than in Apple’s approach, and so is Sami, but Samsung’s approach is integrated to their overall scope of medical devices. Samsung has been working on medical devices already for a long time and is in fact in the same league e.g. as Philips, Siemens and GE. Microsoft’s Health- Vault is closer to HealthKit but the ambition is broader obviously since the Microsoft Health unit has been around for quite a while. Microsoft appears to comprehend health records better than Apple, and Microsoft indeed is operating systems and computer languages more than as compared to Apple, and more than Samsung.
Another aspect here is the emerging mHealth market, where mobile medical devices and solutions will eventually need to go down the same path of approvals (FDA) as compared to other medical devices and drugs. The approval procedures are, however, yet to be defined. The distinction between wearable and nonwearable will be important, as compared to being obtrusive or unobtrusive. In situations involving clinical situations all these platforms run into difficulties as the clinical side defends the professional view on managing health and medical data. However, inclusion of healthy lifestyle interlinked with disease management will promote further use of approaches by Apple, Google, Microsoft and Samsung.
The question is obviously how the market responds concerning interlinks and as related to maintaining integrity with the health records. Qualified-Self solutions still overlook and neglect nomenclature and ontology.
Within analytics and computation, evidence-based medicine (EBM) uses the notation and language of probability and statistics in order to analyze observation of outcomes of individuals in need of care, where we expose individuals to certain treatments or contexts. Logic-based medicine41 (LBM) explains how terminology and nomenclature in medicine can be logically formulated by means of underlying signatures, which in turn leads to the possibility to construct formal terms and sentences, and as they eventually appear within reasoning mechanisms. LBM thereby opens up a logic foundation of probability theory, where notions in probability theory and statistics are enriched with concept used in formal logic.
Logic, as a structure, contains signatures, terms, sentences, theoremata (as structured sets of sentences, or ‘structured premises’), entailments, algebras, satisfactions, axioms, theories and proof calculi (Eklund et al., 2014). Lative logic produces a huge potential of applications using terminology, nomenclature and ontology in particular in social and healthcare. WHO classifications are logically lative (Eklund, 2016). The reference classifications ICD and ICF then appear in structured relation with each other. Similar transformations can be made for the derived classifications as well as for the related classifications ICPC-2, ICECI, ISO9999, ATC/ DDD and ICNP.
Formal mappings, e.g., between ICD and ICF are rare, and this is mostly due to a lack of understanding terminology and nomenclature as terms in a logic. ATC/DDD for drugs embraces ‘dose’ but not ‘intervention,’ which means that drug-drug interactions are possible to describe whereas drug-condition is more complicated. IHTSDO’s SNOMED CT subdivides concepts within its hierarchy consisting e.g. of clinical findings disorders, body structure, pharmaceutical/biologic product, social context, staging and scales, and qualifier values, but has been developed only with intuitive connections with WHO classifications. Further, SNOMED’s assumption that “health ontology” needs the same or a similar underlying logic as web ontology, is a fatal mistake not promoting the dialog and interrelation of classifications and nomenclature in useful application oriented directions. It is also all too narrow to assume that description logic will suffice as a logic for health ontology even if it is suggested to support web ontology.
All this information management is then not just about data and information but indeed about information and process. Information as structure and logic is nomenclature and ontology based, and processes similarly require language far beyond just drawing circles and arrows. Specifically, encoding processes in a more formal and logical manner will need to make use of modeling standards like UML, SysML and BPMN.
For prediction purposes, there is a distinction to be made between a computational algorithm, which includes pattern recognition, neural and Bayesian networks, and similar computational/numerical methods, and logical algorithm, the latter involving inference mechanisms for reasoning is some selected logic, where sentences and statements are based on terms which in turn are founded on nomenclature and ontology. Logical inference manages uncertainty in a different way as compared to computational methods (Eklund et al., 2016). Type theoretical innovations are needed because nomenclature constructions appear in logic as a natural ingredient, but not per se in intelligent computing. A condition is also more of a matter of truth than just a matter of value.
REFERENCES
Eklund, P. Assessment Scales and Consensus Guidelines Encoded in Formal Logic, Journal of Nutrition, Health and Aging (19th IAGG World Congress of Gerontology and Geriatrics, Paris, 2009). 13 Suppl 1, S558-S559.
Eklund, P. Lative Logic Accommodating the WHO Family of International Classifications.
Eklund, P., Galan, M. A., Helgesson, R., Kortelainen, J. Fuzzy Terms, Fuzzy Sets and Systems 2014, 256, 211-235.
Eklund, P., Hohle, U., Kortelainen, J. Non-Commutative Quantales forMany-Valuedness in Applications, in preparation.
Encyclopedia of E-Health and Telemedicine (Eds. Maria Manuela Cruz-Cunha, Isabel Miranda), IGI Global, 2016 (in print). URL: http://www.igi-global.com/book/ency- clopedia-health-telemedicine/141916.
