Consultants - Building a public health intelligence taxonomy and ontology

Purpose of consultancy

The Information Systems and Data Management team is seeking to improve the identification, classification and contextualization of content for the early detection, verification and monitoring of potential health risks in the context of the Epidemic Intelligence from Open Sources (EIOS) Initiative to improve the system and support expansion to Member States. The team is seeking a consultant to work with collaborators to contribute to the development of 4 key outputs as identified in the background below.

Background

The WHO Health Emergencies Programme (WHE) is mandated to enhance WHO’s capacity for the rapid detection, assessment and follow up of all public health risks. This provides the opportunity to establish a framework that facilitates the implementation of a common approach to public health intelligence and event management across all levels of the organization.

The Epidemic Intelligence from Open Sources (EIOS) initiative is a network of organizations with one common goal: to minimize the impact of emerging threats to human health through quality and timely intelligence at the member state, regional and global levels. In doing so, it helps strengthen member state capacity for early detection and evidence-based decision making, aligning with IHR event-based surveillance capacity requirements. The EIOS Initiative is being led by the Information Systems and Data Management team at headquarters, in collaboration with various stakeholders.

The EIOS system uses a broad range of publicly available information to retrieve content which is categorized based on manually pre-defined rules for keyword combinations and exclusions. These category definitions directly affect the quality of the system, impacting the level of noise, the content seen by system users, how content is identified and the ability of users to build relevant and specific filters for the content most relevant to them. Rule-based text classification such as the one used by the EIOS system is understandable by humans and can be improved over time. But this approach has some disadvantages: updates are time-consuming and generating the category definitions is challenging, often requiring a lot of analysis and testing by experts who have deep domain knowledge. Rule-based text classification is also difficult to maintain and does not scale well because adding new rules can unpredictably affect the results obtained using pre-existing rules. Furthermore, category definitions are based on existing knowledge and words and phrases used to build these categories are currently not linked to one another and to their variants as used in the different types of monitored sources (e.g., traditional and social media, official government sources, etc.). No fit-for-purpose public health taxonomy exists, nor is there currently a mechanism to automate the detection of new content, context and linkages that may be of interest beyond the current category definitions.

The absence of a standard public health taxonomy and ontology currently represents a major obstacle to public health activities and has led to a fragmentation within the public health intelligence community. Different organizations are forced to either create their own reference(s) or adopt and adapt existing classification systems designed for other purposes, such as clinical medicine. The development of a global, standardized public health taxonomy and ontology is long overdue. While the targeted outcome is the facilitation of the timely and accurate detection, identification, verification, assessment, and communication of health threats, such a taxonomy and ontology have tremendous benefit to the public health community at large.

The work to be carried out under this contract will contribute to the development of an initial taxonomy and ontology focused on priority areas identified by the team. This includes working with stakeholders to design graph projections and define ways to augment graph data science with automated, machine-learning algorithms for pattern recognition, clustering and anomaly detection.

Operating under an overarching open world assumption, this work will contribute to the generation of four key outputs:

1. Reference public health taxonomy – hierarchical classification of events/entities (all-hazards) based on or adapted from existing taxonomies where available, but specific to public health (this requires collaboration with subject matter experts).
2. Reference public health ontology – canonical, comprehensive public health knowledge representation and reasoning model, consisting of coherent definitions of entities as well as complex relationships between
3. Semantic web of public health intelligence data – dynamically evolving network of distributed, semantically linked data relevant for public health functions stored in different public health intelligence systems as well as in diverse other information systems and datasets that provide relevant contextual information
4. Multi-lingual knowledge representation and language-agnostic reasoning – knowledge representation model implemented as a semantic network of terms in different languages and a reasoning component for language-agnostic computational learning

Deliverables

Outputs and deliverables as identified below are in the context of public health intelligence and will be focused on the priority area of infectious respiratory diseases and related relevant information in English to begin. This will eventually be expanded to other languages and health threats based on findings, changes in the landscape, time and funding.

Output 1: Reference Taxonomy

Activities 1.1 Environmental scan documenting existing taxonomies and ontologies, as well as tool options, and articulating the purpose and need 1.2 Selection of appropriate tool and set up as required within the EIOS infrastructure 1.3 Selection of existing and relevant taxonomies and ontologies 1.4 Development of a vocabulary of terms, definitions and a taxonomy backbone forming the new reference public health ontology in collaboration with communities of experts 1.5 Documentation of all work

Output 2: Reference Ontology and Semantic Web

Activities 2.1 Development, adaptation and/or adoption of a top-level formal ontology, a common semantic data model for selected domain ontologies integrated into the new reference public health ontology 2.2 Design and development of application ontology as an EIOS knowledge graph and integration with reference public health ontology 2.3 Development of recommendations/ standards for a public health intelligence semantic web, including the reference public health ontology and all its parts – top-level formal ontology, vocabularies and taxonomy, domain ontologies, application ontology and communication protocol architecture 2.4 Support development of semantic web API for prioritised PHI systems and contextual information systems

The consultant will be actively engaged on all outputs identified in the Background as they are further explored by the team and other collaborators.

Qualifications, experience, skills and languages

Educational Qualifications

Advanced degree (Master’s or higher) in systematics/taxonomy, health informatics, data science, bioinformatics, applied mathematics, computer science, computational linguistics, computational epidemiology or other related technical discipline(s).

Experience

Essential: At least 10 years of relevant experience in classification and knowledge modelling, including the organization and management of large complex datasets, with at least 3 years of experience in graph modelling.

Desirable: Experience developing and applying models for knowledge management for infectious diseases.

Skills/Knowledge

• Proficiency in Protégé or other ontology editors / knowledge management systems. • Proficiency with graph databases (preferably Neo4j). • Demonstrated ability to interact in complex situations involving technical, cultural and political elements. • Sound analytical and organizational skills.

Languages and level required

Essential: Expert knowledge of English Desirable: Intermediate knowledge of another UN language

Location

Off-site The consultant is expected to liaise closely with the Information Systems and Data Management team in Geneva and other stakeholders, which will require the consultant to be reachable during normal business hours Geneva time.

Travel (If travel is involved, a medical certificate of fitness for work will be required.)

The consultant may be required to travel to meetings with stakeholders and collaborators.

Remuneration and budget (travel costs excluded)

Remuneration: Band level C USD 10,000 - 12,500 per month

Living expenses (A living expense is payable to on-site consultants who are internationally recruited):

Not applicable

Work schedule (if applicable) / Duration of contract:

Start date: As soon as possible End date: Duration of 11 months with up to 11 additional months subject to funding and satisfactory performance

Additional Information section:

• This vacancy notice may be used to identify candidates for other similar consultancies at the same level.
• Only candidates under serious consideration will be contacted.
• A written test may be used as a form of screening.
• If your candidature is retained for interview, you will be required to provide, in advance, a scanned copy of the degree(s)/diploma(s)/certificate(s) required for this position. WHO only considers higher educational qualifications obtained from an institution accredited/recognized in the World Higher Education Database (WHED), a list updated by the International Association of Universities (IAU)/United Nations Educational, Scientific and Cultural Organization (UNESCO). The list can be accessed through the link: http://www.whed.net/. Some professional certificates may not appear in the WHED and will require individual review.
• For information on WHO's operations please visit: http://www.who.int.
• WHO is committed to workforce diversity.
• WHO has a smoke-free environment and does not recruit smokers or users of any form of tobacco.
• Applications from women and from nationals of non and underrepresented Member States are particularly encouraged.
• WHO's workforce adheres to the WHO Values Charter and is committed to put the WHO Values into practice.
• Consultants shall perform the work as independent contractors in a personal capacity, and not as a representative of any entity or authority. The execution of the work under a consultant contract does not create an employer/employee relationship between WHO and the Consultant.
• WHO shall have no responsibility whatsoever for any taxes, duties, social security contributions or other contributions payable by the Consultant. The Consultant shall be solely responsible for withholding and paying any taxes, duties, social security contributions and any other contributions which are applicable to the Consultant in in each location/jurisdiction in which the work hereunder is performed, and the Consultant shall not be entitled to any reimbursement thereof by WHO.
• Consultants working in Switzerland must register with the applicable Swiss cantonal tax authorities and social security authorities, within the prescribed timeframes (Guidelines issued by the Swiss Mission are available at: https://www.eda.admin.ch/missions/mission-onu-geneve/en/home/manual-regime-privileges-and-immunities/introduction/Manuel-personnes-sans-privileges-et-immunites-carte-H/Non fonctionnaires et stagiaires.html