University of Surrey

Test tubes in the lab Research in the ATI Dance Research

Monte Carlo applications for local treatment healthcare usage simulations.

Mee, Thomas (2015) Monte Carlo applications for local treatment healthcare usage simulations. Doctoral thesis, University of Surrey.

Thomas Mee 1552643 Thesis.pdf - Thesis (version of record)
Available under License Creative Commons Attribution Non-commercial Share Alike.

Download (4MB) | Preview
[img] Text
Thomas_Mee_Author_Deposit_Agreement.docx - Supplemental Material
Available under License Creative Commons Attribution Non-commercial Share Alike.

Download (41kB)
[img] Text
Thomas_Mee_RestrictingAccessThesisForm.docx - Restricting access form
Restricted to Repository staff only
Available under License Creative Commons Attribution Non-commercial Share Alike.

Download (579kB)


Cancer is currently the second leading cause of death within the UK. In 2010, cancer caused 29% of all registered deaths, second only to circulatory disease. Cancer incidence is much higher in the elderly than in younger people, so cancer is considered a disease of the elderly. This combined with the fact that the population of England is getting both larger and older suggests that the number of incidences within England will keep on increasing year on year. However, there is an uneven spread of incidences throughout England, different regions have different cancer profiles depending on factors such as age profiles within the region or deprivation levels. There are three principal modalities for cancer treatment: radiotherapy, chemotherapy and surgery. These are used standalone or in conjunction with each other. This project will only look at the provision of radiotherapy. Currently 40% of all patients cured of cancer receive radiotherapy. However, a 2007 study comparing modelled demand and audited activity has indicated that England is currently under providing radiotherapy. The National Cancer Action Team commissioned The MALTHUS project in 2010 after the 2007 study was criticised for being too hard to apply to regional areas, as it is a national model. MALTHUS builds on the 2007 study by creating a piece of demand prediction software that will be used to predict the number of radiotherapy fractions, the appropriate rate of radiotherapy required to be given to England, or within a health care boundary in England. This enables localised models and not just a single, countrywide average. A team of clinicians at Addenbrooke’s Hospital created the clinical decision trees used within MALTHUS and they gained general clinical consensus throughout England. MALTHUS will also utilise high quality incidence and population data to operate a novel Virtual Patient style of decision tree simulation, placing emphasis on the patient walking through the tree instead of the decision tree itself. The decision trees used in MALTHUS are different in design from those created by other research groups. The trees from two major research groups, CCORE and QCRI, use a program called TreeAge and are set up for a single decision with ultimately one of two terminal decisions, radiotherapy or no radiotherapy. The trees in MALTHUS are much more complex with more information within them and give the number of fractions, not just whether radiotherapy is given or not. The basis of them is a Virtual Patient construct where information passes between the tree and the Virtual Patients. It is the Virtual Patients themselves that have the information about the treatment stored and information gathered from all stages of the walkthrough, not just relying on the state of a terminal decision node. MALTHUS Pro also has the ability to offer a suite of statistical and sensitivity analysis options. In terms of tree size, there are 415 branches in CCORE’s trees, but around 2000 nodes in the MALTHUS trees. There are fewer trees from QCRI and so fewer branches. After the successful completion of model verification tests, a complete one way sensitivity analysis on the input data and clinical decision trees was undertaken, highlighting the effect that these parameters have on the outputs of the model. The most sensitive decision out of all cancers is colon cancer - palliative radiotherapy indications or no radiotherapy indications. For each percentage change in the decision the overall answer, for total amount of radiotherapy to be given to a region, changes approximately -1.7% or +2%. Simulations have been completed on both the full Monte Carlo and fast Monte Carlo operating methods, both utilising Monte Carlo integrations, providing a full range of results including the appropriate rate of radiotherapy and number of fractions for every Primary Care Trust and Cancer iii Network within England. Summary tables of these results once again showed how different regions within England are when it comes to the amount of radiotherapy that should be given. The largest difference being in prostate cancer with Hartlepool giving 845 fractions per 100,000 population and Dorset giving 2346 fractions per 100,000 population. The application of deprivation indices show any potential links between the differences in the predicted number of fractions and levels of deprivation. Breast cancer shows a strong correlation to the more deprived while stomach shows a strong opposite correlation. MALTHUS has successfully achieved the original objectives of creating a robust, local radiotherapy utilisation model, using actual population and incidence data to be able to build a true representative cohort of patients and then be able to walk them through a clinical decision tree to predict the number of required fractions. MALTHUS not only works on a regional level and country level, but on any level that the input population and incidence data is available. Additionally the fast Monte Carlo results are applicable to any population/incidence data, including any future predictions in the data, adding in extra flexibility.

Item Type: Thesis (Doctoral)
Divisions : Theses
Authors :
Mee, Thomast.mee1986@gmail.comUNSPECIFIED
Date : 31 July 2015
Funders : National Cancer Action Team
Contributors :
Thesis supervisorKirkby, NFnorman.kirkby@gmail.comUNSPECIFIED
Thesis supervisorJena, Rrjena@nhs.netUNSPECIFIED
Depositing User : Thomas Mee
Date Deposited : 08 Sep 2015 09:27
Last Modified : 31 Jul 2016 01:08

Actions (login required)

View Item View Item


Downloads per month over past year

Information about this web site

© The University of Surrey, Guildford, Surrey, GU2 7XH, United Kingdom.
+44 (0)1483 300800