Olive:
Multi-Agent Human-Computer
Virtual Consultation Engine

Intro:

For making a diagnosis we usually consult a doctor, who helps us determine a cause of a disease and advises the best treatment. Nowadays technologies enrich our lives with new possibilities. Technologies help us in work, in communication, and, of course, they enable us to live healthier lives.

In health-related areas, there is a relatively modern approach called telemedicine, which applies telecommunication technologies for diagnosing and patient care when provider and client are located in different corners of the globe. This approach is rather popular for one-to-one consultation between expert and patient, but it has not been extensively used for group consultations, which bring together the power of many professional doctors from different parts of the planet.

By using technologies from the area of Artificial Intelligence such as distributed human-computer collective reasoning with ontological blackboard, collaborative filtering and case-based reasoning we believe we can unleash the power of expert community by making distributed consultations a reality.

Problem:

Telemedicine could be used to help cut waiting lists and speed up access to distant specialists. However, group consultations carry out even more power - and that is where the product we present, Olive, comes in.

Imagine a world where you can consult with your virtual private doctor using a computer on daily basis. Just imagine the power of professionals form different parts of the world, if they could diagnose a patient together. We aim to make such a technology a reality - which revives a profession of family doctor, combining knowledge from the whole experts community and getting opinions from specialists from all over the world.

Another challenge that we are facing is to reuse the knowledge involved in medical diagnosis, combining the expertise of professionals on the time scale. Medicine is the most ancient of sciences, with a plenty of works made by great medical scientists and the experience of the medicine is so broad, that no-one on Earth could keep it in mind - no-one but computer. The process of experience accumulation can be sufficiently automated, giving us an opportunity to help all parties involved in this process and to make diagnosing more reliable. While some experience accumulation happens on the Internet nowadays, what we present is a formalized approach, making it possible to reuse accumulated knowledge in semantically adequate manner.

Approach:

The goal of our research was to develop a product able to make distributed telemedicine consultations safe, useful and really effective. We wanted to give everybody an opportunity to get virtual consultations with a consilium of professionals without leaving his home - which is a huge advantage for those with disabilities.

We use ontologies as a way to provide formalized communications between personal assistant agents of patients and medical professionals. We annotate doctors' profile by creating associations between his unique identifier and a set of ontology nodes - which is described by logical expressions in description logic. A set of patient symptoms is also annotated using the general Olive ontology, which is derived from the International Statistical Classification of Diseases and Related Health Problems (ICD-10), released and supported by the World Health Organization (WHO).

The concept of ontological relevancy makes it possible to find a doctor the patient exactly needs. Ontological relevancy introduces a metrics between description logic terms relative to nodes in ontology graph, based on the distance in the subsumption tree between specific nodes. The total value is obtained as a weighted sum of distances between pairs of nodes, calculated along the path including their most common subsumer. It is used to determine how relevant a set of symptoms and doctor's profile are.

Although many consultation problems can only be solved by using human experience, it is possible to augment the experts human reasoning with the conclusions of HealthWare - a component of Olive, which is a complex distributed reasoning system that utilizes collaborative filtering, case-based reasoning, ontology relevancy and rule-based reasoning to help experts take the decisions.

Distributed reasoning is based on centralized ontological blackboard that represents the environment for several assistant agents taking part in problem solving. The role of a blackboard is to make agents accumulate and exchange their knowledge in a formalized form. Automated approach allows Olive to collect consultation statistics and provide experts with the possible solution using the methods of collaborative filtering. Experts are also take part in the distributed reasoning process, thus enabling human-computer approach to solution and increasing the synergistic effect of social group during the consultation.

MaiLabs © 2006.
All rights reserved.
Designed by George K,
koo.