My subscription version of Chat GPT ($20/mo) recently got a new function called "Deep Research" (an option on the left-hand menu.)
This week, I asked it to write a report on value-based cancer care, what has held back the field, and whether new kinds of metrics (maybe using AI) could have a big impact.
Chat GPT thought for 45 minutes, and wrote a report (17-page PDF) with some 30 specific footnotes, divided into substantial and content-heavy sections. There are also some tables and graphics.
How good is it? Well, it's pretty interesting and could, at a minimum, serve as start-up orientation to someone who wanted to dig into this topic.
- Here's a cloud copy of the PDF.
- https://drive.google.com/file/d/1qp20YcKG5_y1ddFwnbYkAVixGchmEJit/view?usp=sharing
- Here's the whole report as a very long sidebar blog:
- https://brucedocumentblog.blogspot.com/2026/04/chat-gpt-20-page-report-on-value-in.html
- ##
- And this link, if it works, should take you to an archive copy of the report held in "share" mode at Chat GPT;
- https://chatgpt.com/share/69d559bc-2668-83e8-b253-c591ab138490
- ###
##
AI summarizes its own report -
This 18-page report is a policy brief, research synthesis, and conference-planning memo rolled into one.
Structurally, it moves from an executive summary to sections on why VBCC measurement has underperformed, what measures would better support VBCC, what is lacking today, a future-state vision to 2030, and then a staged roadmap with milestones, governance needs, suggested panelists, and decision-point questions. It also includes a comparative candidate-measures table, a timeline, a flowchart, and a numbered bibliography grounded mainly in CMS/CMMI, ONC, NCI, HL7/mCODE, and peer-reviewed oncology outcomes literature.
The core takeaway is that value-based cancer care has not mainly stalled because of weak payment experiments, but because the measurement system is too claims-centric, process-heavy, fragmented, and poorly tied to patient-centered oncology outcomes. The report argues for a smaller core set of 8–12 digitally computable, equity-stratified measures, especially around ePRO symptom control, function, appropriateness, timeliness, end-of-life care, financial toxicity, and HRSN/equity supports.
As an AI artifact, it is a strong example of “Deep Research Writing Mode”: not just summarizing sources, but assembling them into a decision-ready framework with assumptions, analytic scaffolding, implementation logic, and concrete next-step questions.
In that sense, it reads less like an essay and more like an AI-authored strategic operating document.
