Feasibility Study: Integrating Games-Based Learning and Computational Modelling to Control MRSA.

Lead Research Organisation: University of Abertay Dundee
Department Name: Sch of Computing and Creative Tech

Abstract

MRSA is a 'super-bug' that is difficult to control using antibiotics, and is most common in hospitals. Around 1 in 10 people admitted to hospital will contract an infection during their stay. You can become infected through physical contact with either another infected person or a surface like a door handle already touched by an infected person. If you are really ill, as many people are in hospital, the infection can kill you. There are ways to limit the spread of MRSA in hospitals, including healthcare workers following strict hygiene measures such as washing their hands between patients, isolating infected patients and even closing down hospital wards where infections occur. The problem is that each option can cost a lot of money which could otherwise be used to treat patients and nobody knows the best way to use these different options together to manage an outbreak of MRSA.Researchers have turned to computer modelling to help them understand how to manage the complicated range of factors involved in spreading the infection. The models show that the pattern of movements of healthcare workers among patients is really important and if the activities of healthcare workers are managed properly then spread can be limited. Also crucial is the degree to which healthcare workers follow the hygiene measures. Effective training of healthcare workers on the importance of following the hygiene rules can also limit spread. Most important of all, the models show that each individual person involved can make a big difference to the occurrence and spread of the disease. To decide how to manage the spread of infection in a particular ward, you need to know about the ward layout and the people that work and are being treated in that ward. Unfortunately, none of the models cater for differences in the behaviour of individual healthcare workers and the health of individual patients. Also, these models do not represent the layout of the ward, and how healthcare workers move around in the ward itself. We have developed a computer model that takes into account both the layout and the individuals in a hospital ward. We will add data about healthcare worker behaviour, healthcare activities among patients and individual patient health from a special study ward in a hospital into the model. We will also include data from different studies on the different ways to manage the spread of MRSA. We will combine our MRSA spread model with an existing training tool to teach healthcare workers about the importance of following the hygiene rules. We will clearly demonstrate to them how many patients one careless person can infect, and how careful they need to be to help reduce infections. This training tool uses a computer game approach to provide an interesting way of teaching.Since the computer model predicts realistic outcomes, it can also be used by managers to choose the best method of containing an ongoing outbreak. We will use artificially intelligent search techniques to identify the best way of combining different approaches to contain the spread. The same approach can also be used by managers to identify ways of reducing infections in the first place, and to plan ahead by getting the computer to simulate different possible scenarios and to identify ways of dealing with them. The system that we are aiming to build will help hospitals manage the spread of MRSA in a cost-effective way. It will show hospital managers possible ways of limiting spread in their hospital, and teach healthcare workers about the difference they can make in reducing the chance of an outbreak.

Publications

10 25 50
 
Description MRSA is an infection that can be spread through contact direct between an infected and an uninfected person, or through indirect contact through touching for example a shared surface. In a hospital setting care workers tend to patients, and both patients and carers mix amongst themselves. Thus patients can become infected with MRSA as a result of patient-patient contact, patient-carer contact and patient-environment contact.

The project developed a proof-of-concept model and accompanying visualisation to show that agent-based modelling can be used to describe MRSA spread. Agent-based modelling is an approach used to describe systems characterised by agents interacting in some space over time. Here, our agents were the patients and carers, and our space was the hospital ward. We modelled patients being tended to by carers, with spaces for carers to mix and spaces for patients to mix.

The model allowed us to specify the likelihood of infection transfer for each interaction through probabilities. A key strength of agent-based modelling is that you can vary the properties of the agents, for example carer-dependency levels of patients.

In this short project we calibrated the model using a mix of literature and expert input from hospital infection control specialists. We demonstrated that when we varied the parameter settings of rates of transfer and agent properties we obtained sets of results that were plausible. This gave us confidence that this modelling approach could be used as a framework for understanding the spread of infection control, and so how to control that spread, in the hospital ward.
Exploitation Route This was a clear demonstration of the potential for agent used modelling approaches to be used in disease spread management, especially in systems where variation in individual behaviour and the utilisation of physical space is important.
Sectors Healthcare

 
Description This award was to pump-prime the use of agent-based computational modelling approaches developed in ecology to model the spread of MRSA in the hospital ward. The project demonstrated sufficient proof-of-principle to support a successful application to TSB for a Knowledge Transfer Partnership with NHS Tayside (£240K awarded in 2006).
First Year Of Impact 2006
Sector Healthcare
Impact Types Policy & public services

 
Description Ninewells Hospital & Medical School 
Organisation Ninewells Hospital
Country United Kingdom 
Sector Hospitals 
PI Contribution We provided the computational model of MRSA spread, including patient agents, carer agents and their interactions, together with the accompanying user front end design and implementation to depict the model dynamics in a user friendly manner.
Collaborator Contribution The partner provided all domain expertise to enable us to construct the model. This input included overall design issues such as the ratio of carers to patients, shift patterns, and agent characteristics with parameter values for each. Specified parameter values included rates of transmission, details of patient and carer behaviours and patterns of patient-carer, intra-patient and intra-carer mixing. They also provided user feedback on the interface design.
Impact Outcome: 2-year KTP partnership Output: Milazzo, L., Bown, J. L., Eberst, A., Phillips, G., Crawford, J. W. 2011. Modelling of Healthcare Associated Infections: A study on the dynamics of pathogen transmission by using an individual-based approach. Computer Methods And Programs In Biomedicine. Vl 104. Is 2. 260-265.
Start Year 2006
 
Description Team Play Learning Dynamics 
Organisation Team Play Learning Dynamics
Country United Kingdom 
Sector Private 
PI Contribution We provided the computational model of MRSA spread, including patient agents, carer agents and their interactions, together with the accompanying user front end design and implementation to depict the model dynamics in a user friendly manner.
Collaborator Contribution The partner provided games-based learning expertise to inform the development of the user interface to the computational model to ensure that interface was an effective communication tool.
Impact None
Start Year 2006