Skip to main content
Get your brand new Wikispaces Classroom now
and do "back to school" in style.
Data Science | DSDHT
Are there any new drugs left?
Boiling the ocean - integrating everything for patient care
Boiling the ocean - integrating everything together
Gene expression data analysis
Information-based Drug Discovery
Introduction to R Shiny
Managing disease with data science
Mapping Structure to function
Add "All Pages"
Managing disease with data science
In the introductory video, we looked at three supplemental questions that we need to ask, when we shift to a person-centric health model:
How do I stay as fit and healthy as possible?
How do I know when something is going wrong?
How do I respond when something is going wrong?
In this short module, we will be looking at the last two questions, using Diabetes as an example. In the USA,
25.8 million children and adults – 8.3% of the population - suffer from Type II diabetes
. In 2007, it was listed as the underlying cause of death for over 71,000 people, and a contributing factor for a over 160,000 more. Further, people with diabetes suffer from serious complications including heart disease (68%), stroke (16%), high blood pressure (67%), and vision problems (28%). It has been estimated that the cost of diabetes in 2012 was $245 billion ($176 billion in direct
medical costs, and $69 billion in reduced productivity).
Despite this, the disease is poorly understood. Type 2 diabetes is primarily defined by insulin resistance (i.e. the body not producing enough insulin, or not being able to use insulin properly), and is associated with familial history and weight problems. However, the root causes are not understood, and the disease is only poorly understood at the molecular level. Current medications for type-2 diabetes have focused on stimulating the pancreas to produce more insulin, or increasing use of insulin in cells, but such medications can have problematic side-effects (Troglitazone, Rosiglitazone, etc).
So the fist question - how do I know when something is going wrong? - can be framed as a data question: what is the risk, based on the data, of me getting diabetes, and what are the signs in the data of this happening? At a very general level, this can be done using very crude measures, such as BMI, family history, and size. For example, the
Diabetes Risk Calculator
uses these factors to come up with a measure of risk. This can be approached more qualitatively - see for example
Am I at risk for Type 2 Diabetes
So how do I know when something is going wrong? Well, we can look up our symptoms on one of the many established references such as WebMD or
, which map established relationships of symptoms to conditions. A much newer, more data science focused approach (but as yet not really validated) is to do "surveillance" of symptoms and conditions from patients using social media, physician reports, and other data sources. An excellent resource which starts to move towards this approach is
(see for instance their calculators).
So we find out we have type II diabetes. What do we do next? Well there are a
plethora of apps
available to help do things like monitoring of diet and glucose levels on an individual basis.
But start to think about where this could go in the future -- for example, collecting data on relationship of food intake to glucose levels for millions of patients, with information about other conditions and factors - whether they have cardiovascular disease, their weight, BMI, and so on. Or mining social media for similar kinds of trends.
How a person calculated its diabetic chart using some smart apps and devices and analyzed the data
Predictive analysis of diabetic treatment using a regression-based data mining technique for old men and young students
Managing diabetes with Data Science
Personal health data analysis - Diabetes Experiment
help on how to format text
Turn off "Getting Started"