Evaluation of Machine Learning Algorithms for Patient Experience Survey Data Analysis

  • Mohammed AlQedrah

    Student thesis: Master's Thesis


    The growth of machine learning techniques in the past few years caused it to be the interest of many industries, and a major industry looking to benefit from advances in machine learning is the healthcare industry. Improving quality of services provided by health care organizations or corporations enhances patient experience, which results in a healthier community in addition to increase in revenue for such organizations; achieving this goal could be facilitated machine learning. This research dives into the analysis of an inpatient experience survey provided by the NHS which includes 20 independent variables (grouped under 5 domains) and one dependent variable (overall patient experience) using a total of seven different machine learning algorithms. Moving on, some machine learning algorithms are frequently used by survey analysts. On the other hand, the remaining five machine learning algorithms are algorithms that have potential in survey analysis. Overall, all machine learning algorithms used in this research performed extremely well. However, linear based machine learning algorithms showed better results. Nonetheless, linear based machine learning algorithms had an advantage in this dataset as the dataset was extremely linear in nature. Feature importance is also a crucial aspect of this research. Hence, an analysis using the different machine learning algorithms available showed that domains representing aspects such as 'Building closer relationships with patients' and 'Better Information, providing more choice for the patients' were the most influential when it came to overall patient experience. Hence, individual question belonging to such domains such as 'Were you involved as much as you wanted to be in decisions made about your care and treatment?'. On the other hand, the lowest scoring individual question 'Was your admission date changed by the hospital?' belonged to the least correlated domain which represented the theme of 'Access and waiting'.
    Date of AwardMay 2021
    Original languageAmerican English


    • Machine Learning
    • Healthcare Industry
    • Data Analytics
    • Random Forest.

    Cite this