I am Prachi Bagave, a first year PhD student at the Delft University of Technology, and my research focuses on developing accountable AI solutions for healthcare IoT systems. It is funded by the H2020 SPATIAL project, and fits quite well under the broad trustworthiness framework of the project.
My PhD is also uniquely positioned at the faculty of Technology, Policy, and Management, a multidisciplinary faculty that enables me to look at the accountability problem from a socio-technical lens. On the other hand, the SPATIAL consortium consists of four universities and eight SMEs, each focussing on one or more trustworthiness elements such as privacy, resilience, security, biases, and accountability. Thus, we fit the different pieces together to make a complete trustworthiness framework. We also work on various use cases to realize these elements and bring them into practice.
The focus of my study is to develop design elements for accountability in the remote monitoring systems of the healthcare IoT domain, and I do so by exploring the needed features from an interdisciplinary study and exploiting XAI methods to operationalize them. I take a socio-technical approach to understand the underlying issues and develop methods to make the AI–user decision-making more accountable. Moreover, I take a more pragmatic approach by using causal inferences to enhance the interpretability of AI decision-making. In my research, I also shed some light on how this can help us make more trustworthy relationships between AI and human users.
Reflecting on last year, my journey started when I briefly met Aaron Ding, the TUD team, and the Erasmus University team during the project kick off day, when I had not officially started with my PhD position.
My first impression of the team was quite friendly, and in addition, I was also quite fascinated by the work explained to me during the recruitment. I have come a long way since then. My purely technical approach from a data and system standpoint has been guided by the various social talks I had during my work. I attended a summer school to explore my eXplainable AI knowledge, two project consortium meetings where we brainstormed on our project platform ideas, and in the near future already lies two planned conferences where I publish and discuss my work with other researchers.
We also included a new member, Marcus Westberg, in our TUD team, said some goodbyes, and welcomed new members in the consortium at the different organizations. Overall, it has been a gradual learning curve so far, and I envision having a much more fascinating journey ahead, meeting new people, exploring new ideas, and developing trustworthy solutions on the way forward.