BACKGROUND
Diabetes mellitus is an increasingly common life-style disorder whose management outcomes are measured in symptomatic, biochemical as well as psychological areas. Well being as an outcome of treatment is being increasingly recognized as a crucial component of treatment. There is little published literature on psychosocial outcomes and the factors influencing them. Therefore we have developed a neural network system which is trained to predict the well being in diabetes, using data generated in real life.
MATERIAL AND METHODS
We developed a Multi Layer Perceptron Neural Network model, which had been trained by back propagation algorithm. Data was used from a cohort of 241 individuals with diabetes. We used age, gender, weight, fasting plasma glucose as a set of inputs and predicted measures of well-being (depression, anxiety, energy and positive well-being).
RESULTS
It was observed that female patients report significantly higher levels of depression than their male counter parts. Some slight high or no significant differences are observed between males and female patients with regard to the number of persons with whom they share their anxieties and fears regarding diabetes. There is not much difference has been observed in energy levels of both males and females. Also, Males have higher pwb value when compared with the female counterparts. Also, this may be due to women tend to react more emotionally to disease and hence experience more difficulty in coping with it. The present sample of women being predominantly house wives may be worrying more about their health and its problems. Also, it is observed that, gender differences are significant with regard to total general well-being. With five inputs (age, sex, weight, fasting plasma glucose, bias), four outputs are four (depression, anxiety, energy and positive well-being) the momentum rate was 0.9, the learning rate 0.7, using a sample of 50. the maximum individual error is 0.001 when the number of iterations were 500, number of hidden layers is 1 and the number of units in the hidden layer are 6, the normalized system error was 470.57. With input samples of 100, 150 and 200, keeping the other variables constant, the normalized system error was 419.61, 359.67 and 332.32 respectively. Similar values are found for the normalized system error when the number of units in the hidden layer have been increased to 7, 8 and 9 respectively. With two hidden layers, and with each hidden layer containing 6,7 ,8, 9, 10, 11 units for the samples 50, 100, 150, and 200, the same values of normalized system error was found. Women having weight between 40 kgs and 85kgs had higher levels of depression than men who had weight between 39kgs and 102 kgs.
CONCLUSION
We have developed a prototype neural network model to predict the psychosocial well-being in diabetes, when biological or biographical variables are given as inputs. When greater data was fed to the system, the normalized system error can be reduced.
[1]
Clare Bradley,et al.
The Well-being Questionnaire.
,
1994
.
[2]
Sumitra Mukherjee,et al.
A Bayesian Framework to Determine Patient Compliance in Glaucoma Cases
,
2000,
AMIA.
[3]
G. Sridhar,et al.
STRESS IN THE CAUSE AND COURSE OF DIABETES
,
2001
.
[4]
S. Zickmund,et al.
Living with Diabetes
,
2008,
The Diabetes educator.
[5]
G. Sridhar,et al.
Living with diabetes: Indian experience
,
2007
.
[6]
Zsolt Tulassay,et al.
Application of neural networks in medicine - a review
,
1998
.
[7]
S. Anitha,et al.
Application of a radial basis function neural network for diagnosis of diabetes mellitus
,
2006
.
[8]
V. Mohan,et al.
Epidemiology of type 2 diabetes: Indian scenario.
,
2007,
The Indian journal of medical research.
[9]
G. Sridhar,et al.
Psychosocial and cultural issues in diabetes mellitus
,
2002
.