How a Data Analyst Treated her Hypertension with Predictive Analysis

0

Keeping Calm During Stress

Translating numbers to plain English and Predictive Analysis was something that Kiara did everyday as a Data Scientist, to help her company make better business decisions.

One day, a simple saliva test revealed that Kiara’s DNA had some markers that made her genetically predisposed to a high risk of Hypertension or High Blood Pressure. Hypertension is a biological trait that is highly inheritable and gets passed on from one generation to another.

Having spent 40 summers on this planet, Kiara’s health was something that she had started to care about more seriously than anything else. Regular Health Check-ups had become one of the most important items on her to-do list. The feeling of anxiety that she would get whenever she visited a hospital was so much that she would end up with a high blood pressure each time, her vital statistics were measured. But the Doctors would always dismiss it as a  ‘White-coat’ hypertension. On one such occasion, Kiara asked the Doctor if she should just go ahead and start taking medicines. But the Doctor advised that the drug could make the blood pressure too low, which is even worse than having hypertension. The Doctor suggested that she buy a digital meter, check her blood pressure at home, and get him some readings before they decided on this issue.

So, Kiara’s stint with the BP meter started with her taking readings every morning and evening. In a span of 6 months, she had collected a whole lot of data comprising date, day, time, systolic blood pressure, diastolic blood pressure and pulse rate. 

Kiara decided to do further analysis on her own clean data that had no missing values and yes, some outliers. She started with some descriptive statistics. She calculated the measures of Central Tendency, measures of variation and measures of shape. She plotted a curve to visualize the distribution of data and ended with a near normal or a Gaussian curve for 40-50 age bracket. She conducted a two-sample test to check if her average was statistically significantly different between two months. She went on to do an ANOVA to check the impact of different factors on her blood pressure.

As her dataset was consistent and well-defined, she decided to do a Time Series Analysis on her data to forecast her blood pressure for the next 6 months. When she decomposed the data, she found some cyclic variation and but not much of an increasing trend and yes, a lot of random fluctuations.

She added more features to her data, like salt intake, exercise in minutes or yoga in minutes, her stress levels or even having guests over and was able to predict which features led to a rise in her blood pressure.  A simple clustering of the data helped her find reasons for different kinds of fluctuations in her blood pressure.

With survival analysis on her data, she was able to able to make a lot of changes to her lifestyle and delay the onset of the dependence on pills to control her blood pressure.