Speaker: Dr. Nguyen The Loc, HUMG
Topic: Non-Pharmacological Interventions (NPIs)
Date: 2:00pm – 3:00pm, October 18th 2018
Location: ICTLab, USTH building, 18 Hoang Quoc Viet, Cau Giay, Hanoi.

Description: “Non Pharmacological Interventions (NPIs) are science-based and non invasive interventions on human health. They aim to prevent, treat, or cure health problems. They may consist in products, methods, programs or services whose contents are known by users. They are linked to biological and/or psychological processes identified in clinical studies. They have a measurable impact on health, quality of life, behavioral and socioeconomic markers. Their implementation requires relational, communicational and ethical skills.” (Plateforme CEPS, 2017).

In recent years, Non-Pharmacological Interventions (NPIs) have attracted a lot of attention in the health care community. NPIs can no longer stop at a professional discipline to describe them (psychotherapy, manual therapy, dietary supplement, adapted physical activity, e-health solution, etc.). It requires access to a more concrete level of description where each NPI can be evaluated by science, monitored by professionals and explained to the patient. To do this, an international and evolutionary classification based on the results of science is necessary. Thus, developing an ontology for NPIs is crucial.

Constructing this ontology manually is time consuming and thus an expensive process. Particularly, the step of collecting the NPI terminology requires much more time than the rest, because of heterogeneous and big resources in the context of NPIs. An automatic or semi-automatic method is thus essential to support NPI experts in this task.

In this talk, we would like to introduce a three-step process (term extraction, term ranking, term validation) to acquire NPI terms. In the first step (term extraction), candidate terms are extracted from ontologies in BioPortal, which is the largest biomedical ontology repository, and from article abstracts in PubMed, which is a large biomedical bibliographic database with more than 28 million citations for biomedical literature. In the next step (term ranking), the candidate terms are first filtered to remove poor candidates, then the NPI score based on semantic similarity is proposed to rank the remaining candidate terms. Finally, in the last step (term validation), the potential NPI terms are validated by NPI experts in order to confirm the new NPI terms.

Keywords: Terminology Extraction, Terminology Acquisition, Semantic Similarity, Non-Pharmacological Interventions.