HCII Ph.D. Thesis Proposal: Megan Hofmann, "The GIST of Medical Making: A Framework For Producing Clinical CAD Tools with Generative Design"

Megan Hofmann

Tuesday, January 12, 2021 - 1:00pm
Zoom meeting (given in email announcement)

Scott E. Hudson (HCII), Co-Chair

Jennifer Mankoff (UW), Co-Chair

Patrick Carrington, (HCII)

Jodi Forlizzi, (HCII)

Bjorn Hartmann, (UC Berkeley

Disabled people and clinicians have a long history of inventing novel assistive technologies for themselves and others. Their practice of creating assistive technology at the point-of-care is necessary to assure access to assistive technology. Disabled people, like all people, should be supported in augmenting their environment to their needs. Further, clinicians, as most people's primary providers of assistive and medical devices, are best situated to deliver customized and personalized care.

It would at first appear that increased access to consumer-grade digital fabrication technologies like 3D printers would increase the scale and scope of their efforts. Unfortunately, the design tools needed to access digital fabrication are designed, primarily, to serve engineers, professional designers or novices with no particular design focus (e.g., children, hobbyists). Such tools only leverage technical expertise, not the rich domain-expertise a disabled people and clinicians could lend to the design effort. A disabled person is uniquely qualified to express their needs and a clinician's medical expertise is often critical for designing assistive and medical devices that are verifiably safe and effective. Unfortunately, there is currently no way to express that expertise in design tools.

I present two relevant bodies of research. The first is two qualitative studies of \textit{medical making}, the practice of designing assistive and medical devices in a clinical setting or for use in a clinical setting. The second is the development of two computer aided design frameworks which support the (1) representation of domain specific expertise and (2) utilization of domain knowledge in generative design.

I focus on medical making, as opposed to general assistive making, because it is both understudied and a critical access point where many people receive assistive devices. My qualitative research reveals that medical makers (e.g., clinicians) are effective at recognizing patient's needs and design requirements but struggle to use standard design tools to actualize  a design that meets those requirements. That is, they can prescribe a solution, but they struggle to make it. To meet the full potential of medical making, domain specific design tools would need to enable designers to specify their requirements, automatically generate designs to meet those requirements, and utilize design patterns from existing devices to inform this generative process.

My two existing frameworks contribute ways for makers to specify their intentions while creating 3D models and to use their intentions to generate designs that meet their specifications. My first framework, PARTs, scaffolds the 3D modeling process like object-oriented programming but in an entirely geometric (non-programmatic environment). This helps designers to add functionality and documentation to their design, amplifying their expertise as designs are shared and reused. My second framework, GIST, is the core, technical contribution of this thesis. GIST structures the creation of domain-specific generative design tools, particularly those needed for medical making, around the mapping of objectives (i.e., designer goals) to tactics (i.e., ways of reaching those goals). While GIST is derived from my design requirements of clinical CAD tools, it is also a generic and extensible framework which can be applied to a wide range of domains. Thus far, I have evaluated GIST by applying it to the novel domain of automatic machine knitting. In my proposed work I will improve GIST by advancing my contributions to machine knitting and applying it to two new assistive domains (e.g., tactile maps, hand-splints). 





Megan Hofmann
Human Computer Interaction Institute
Carnegie Mellon University

Upcoming Events