Decision Support Systems

With ADVANA the scientists at ATIA generate predictive models that form the basis for Decision Support Systems (CDS). For example, the integration of the output of our decision tree generating algorithm Moku and the input of our deductive reasoning agent Ceres allows for very short turnover times in making or updating a CDS. As a bonus, the predictive models themselves offer insights into the variables explaining response as well as non-response. The latter can also be used to identify new targets for treatment approaches.

Clinical Decision Support Systems

However, the proof of the pudding for the predictive models is to convert them into a CDS, that integrates with the Electronic Medical Record to parse the patient characteristics necessary for the given decision and send the advise in return. It is extremely important that the advice is transparent: the clinician needs to know on what information it was based, what the level of (un)certainty is, and why other treatment alternatives were dismissed. This enables the clinician to test his own education-and-experience based preference against the CDS advice. The ATIA analytics are designed to be transparent and support this kind of professional use.

Clinical Decision Support (CDS) systems help health care professionals to make decisions in complex medical problems. The ATIA systems reason with patient knowledge and give the right information, to the right professional, through the right channels at the right time. This:

  • Improves patient safety (e.g. by reducing medication errors through alerts)

  • Improves the quality of care by making available up-to-date medical knowledge, guidelines and protocols that will lead to the correct diagnosis and treatment

  • Improves efficiency and reduces costs.

Speed and specificity

Finally, CDS need to work real time. At ATIA, we have gathered a lot of experience in fast reasoning when developing a medication safety surveillance system, that implemented tens of thousands of rules governing the safe prescription of medication. At the same time such a system should only issue warnings when they are appropriate for the patient at hand, e.g.: warnings about medication that may be harmful to the unborn child should only be given to women whor are or might pregnant, not to men or old age pensioners.

ATIA skills

ATIA has both the AI skills and the (bio)medical knowledge to build such CDS systems.
Our intelligent agents can reason with vast amounts of very diverse patient data to help establish diagnoses, underlying causes and suggest optimal treatment for the patient at hand. Multiple agents interact on an agent platform and are heterogeneous in either the way they reason or the knowledge domain they act in. 
HeMAS stands apart from classical data analysis techniques in computational skills, transparency and performance. ATIA's goal is that CDS system vendors will want nothing else than our HeMAS engine for their products.

ATIA needs to demonstrate that CDS systems using our technology are better than systems that don't use that technology. So far, the end users are thrilled at the results of our implementation.
ATIA's unique technical selling points are:

  • They outperform existing methods, in both specificity and sensitivity

  • They are fast

  • They are reliable

  • They are transparent

To remain ahead of our competition we continuously strive to improve our agents and their co-operation. Therefore, scientific research remains one of the core activities of ATIA.





© 2015 Alan Turing Institute Almere