(147r) Techniques to Assist the Human-in-the-Loop in Type-1 Diabetes and Smart Manufacturing

The focus of our work is on developing algorithms for disturbance detection and designing human-machine interfaces (HMI) to reduce the burden on patients with Type-1 diabetes and assist the workforce in small and medium-scale enterprises.

Type-1 diabetes (T1D) is an autoimmune condition wherein the patient’s pancreas ceases to produce or cannot produce enough insulin to counter the blood glucose. Individuals with T1D have to endure the burden of calculating the impact of multiple disturbances on their blood glucose levels. These disturbances include daily activities like meals, exercise, and sleep, system faults like Pressure Induced Sensor Attenuation (PISA), and infusion set failures. Our work focuses on extending an event detection framework based on particle filtering to include information from observed glucose levels. Efforts have been made to create an iPhone and smartwatch app for gathering accelerometer data and daily logs. The design of the HMI prioritizes user-friendliness and simplicity to streamline the data collection process.

In the case of Smart Manufacturing, we explore the domain in small and medium-sized food processing enterprises, where operators go through manual data and quality analysis. Product consistency is important since the food industry relies heavily on customer satisfaction. The project aimed to digitalize the sensors, quantify product quality, reduce product loss, and upskill the workforce. An integrated package containing data transfer and storage from sensors to the cloud platform, a disturbance detection algorithm, and a user-friendly interface was provided as a solution to the industry. The Smart Manufacturing and Innovation Platform (SMIP), developed by the Clean Energy Smart Manufacturing Innovation Institute (CESMII), was used for seamless data collection, transfer, and digitalizing of the sensors.

Another case study of Smart Manufacturing was studied using a laboratory shell and tube heat exchanger. Our work takes a crucial step toward bridging the industry-academia gap by enabling students to conduct better experiments using an experiment assistant and access to larger datasets to catalyze their interest in developing predictive models for fault detection. Provision is made for students to develop and test their fault-detection algorithms. The designed human-machine interface can be used to study and conduct lab-scale studies to understand human-machine interaction and the Human-in-the-loop aspect of decision-making.

Research Interests: Smart Manufacturing, Industry 4.0, Type-1 diabetes, Human-in-the-loop decision-making, Interface design