Joanna Ng, IBM, #Cascon 2012 workshop on Analytics of Big Data Generated by Healthcare and Personalized Medicine domain.
Process of fixing bugs isn’t enough.
Compare software development to film production.
Start with big data platform, used for health informatics.
Stereotype users, then understand purposes, goals, tasks, interactions.
First problem space, then ideation of shifts, in contrast to engineering approach.
Informatics as analytics, information retrieval, semantic reasoning, data mining.
Data: unstructured, what is trustworthy? Could find some drug more efficacious, since all patients are different.
Predictive analytics: e.g. which chemotherapy will statistically yield the longest survival years, for the physician at time of decision.
Data mining for new patterns on inferences, associations
First challenge: what question should we ask? Doctor can’t look at data when patient is in front of him/her.
Interface as question / answer with natural language, or search query?
Data entitlement: directly manipulate, use, view; expose / exchange data
Functional entitlement: direct or indirect data operations; see or can’t see
Need this understanding to design the interface to achieve goals