BMT 342 Lecture 8 PDF

BMT 342 Data Mining in IoT/IoMT Lecture 8 Dr. Asma Abahussin Department of Biomedical Technology College of Applied Medical Sciences King Saud University 1 Objectives To learn and understand: ▪ What is Data Mining? ▪ The importance of data mining. ▪ Data mi...

BMT 342 Data Mining in IoT/IoMT Lecture 8 Dr. Asma Abahussin Department of Biomedical Technology College of Applied Medical Sciences King Saud University 1 Objectives To learn and understand: ▪ What is Data Mining? ▪ The importance of data mining. ▪ Data mining techniques. 2 Introduction ❖ IoT/IoMT systems generate huge amounts of raw data. ❖ Business owners need to become more data-driven instead of just collecting and storing large amounts of data. ❖ They should extract the maximum value of the data and, based on the insights, make strategic business decisions. ❖ They need to employ data mining! 3 What is Data Mining? ❖ Data mining is a subset of data science that refers to the process of discovering patterns (relationships) and other key information from massive data sets. ❖ It relies on machine learning and statistical analysis. 4  Data Mining Techniques ❖ Various techniques can be used to mine data: 1. Classification: It classifies data into different classes based on important and relevant information about the data. ▪ It is useful in building predictive models. ▪ Examples of application: Disease Diagnosis: Classification can be employed to diagnose diseases by analyzing patient data and identifying which category a new patient’s symptoms fall into based on historical data. Patient Risk Assessment: Using historical patient data to predict their risk of developing certain conditions, such as predicting cardiovascular diseases based on lifestyle data and previous health records. 5 Data Mining Techniques (cont.) 2. Clustering: It groups data elements that share similar characteristics together into clusters. ▪ It is useful in identifying hidden patterns. ▪ Examples of application: Patient Segmentation: Clustering can be used to segment patients based on various characteristics, such as demographics, medical history, or response to treatments. This allows for more effective personalized care plans Disease Patterns: Identifying clusters of similar disease symptoms or progression patterns among patients helps understand disease mechanisms and potential treatment protocols. 6 Data Mining Techniques (cont.) 3. Regression: It finds a relationship between a dependent (target) and independent variable/s (predictor) ▪ It is useful for predicting specific outcomes or defining the probability of a variable based on other factors. ▪ Examples of application: Predicting Treatment Outcomes: Regression can forecast patient outcomes based on treatment plans. They can predict how patients will respond to different treatment options, which helps in customizing patient care. Resource Utilization: Predicting hospital resource needs such as bed occupancy, staffing, or medical supplies based on trends in patient flow and hospitalization rates. 7 Data Mining Techniques (cont.) 4. Outlier Detection: It identifies data points that deviate significantly from the rest of the dataset. ▪ It helps detect irregularities and take possible actions accordingly. ▪ Example of application: Unusual Patient Data Monitoring: Outlier detection can identify unusual patient readings or vital signs that deviate from normal patterns. This can be critical in acute care settings or for monitoring patients with chronic conditions remotely. 8 Are All of the Patterns Interesting? ❖ Data mining may generate thousands of patterns, but not all of them are interesting. ❖ A pattern is interesting if it is: Novel. Potentially useful. Valid on new or test data. ❖ An interesting pattern represents knowledge that can be used to drive decisions! 9