01
The umbrella
Patient-Side
A patient-side health-intelligence framework for carrying lived experience into professional decision settings.
Patient-side health intelligence · working framework
Longitudinal context should survive the handoff.
Patient-Side develops methods for preserving lived experience across time—then carrying it into professional settings without disguising uncertainty, erasing contradiction, or transferring authority to AI.
Preserve provenanceKeep evidence states visibleRetain contradictionReduce human burden
168hours lived
1brief encounter
01 · The failure mode
Compression is inevitable. Integrity loss is not.
Life is continuous; professional understanding is episodic. Meaning can degrade when experience is summarized, inferred, compressed, or handed off. Patient-Side calls that failure Longitudinal Integrity Loss.
The goal is not more data. It is usable context with an honest chain of custody.
02 · The system
One framework. Four working parts.
Patient-Side is the applied health case. Its components separate capture, evaluation, burden, and handoff so each can be tested rather than treated as one black box.
02
The evaluation layer
PSLAI
An audit discipline for testing whether longitudinal AI preserves meaning, evidence state, and human authority.
03
The usability test
Burden Audit
A check on who must capture, correct, interpret, and carry the work—and whether the workflow earns that effort.
04
The bounded output
Clinical Handoff
Purpose-specific material that preserves a way back to sources and presents better questions, not machine diagnoses.
03 · The lifecycle
From observation to accountable update.
Human interpretation does not end the record. It returns to it—creating an accountable longitudinal loop.
- 01
Capture
Record close to the moment.
- 02
Structure
Organize without rewriting.
- 03
Longitudinalize
Place events across time.
- 04
Interpret
Develop bounded candidates.
- 05
Measure
Test recurrence and change.
- 06
Hand off
Present purpose-specific context.
- 07
Human decision
Keep judgment accountable.
- 08
Update
Return interpretation to the record.
04 · PSLAI audit discipline
Test whether meaning survives.
PSLAI evaluates longitudinal record AI by the integrity of what it preserves—not merely the fluency of what it produces.
Three noncompensable gatesProvenance · Evidence-state separation · Human authority & safety
01
Provenance & traceability
02
Chronology
03
Evidence-state separation
04
Contradictions
05
Uncertainty
06
Negative evidence
07
Patient voice
08
Function & burden
09
Cross-specialty context
10
Actionability
05 · Governance at the interface
Formal control is not the same as usable agency.
Substantive Human Agency asks whether a person can actually understand, correct, contest, limit, and direct what an AI system does across institutional boundaries.
Information
What is known, reported, inferred, missing, or contested?
Authority
Who may remember, recommend, communicate, represent, or act?
Accountability
Can errors be found, corrected, challenged, and traced?
Burden
Who does the work—and who benefits from it?
Delegation must not expand authority.Purpose limits, data boundaries, and accountability must survive agent-to-agent handoffs. Shared memory and external artifacts are governed coordination surfaces—not neutral infrastructure.
06 · What can be examined now
A practical wedge: evaluate longitudinal record AI.
The immediate contribution is an evaluation method: inspect where a longitudinal system preserves or loses provenance, chronology, evidence states, contradiction, patient voice, functional meaning, and professional authority.
Map the recordTrace source material through compression and handoff.
Audit the outputScore integrity dimensions and enforce critical gates.
Test the burdenMeasure the work imposed on patients and professionals.
Build the validation pathSeparate demonstrated practice from hypotheses requiring study.
07 · Development status
Clear about what exists—and what does not.
The framework is being formalized through documented practice, audit design, comparative research, and a validation roadmap.
Exists now
A documented framework and audit model
A repeatable lifecycle, named failure modes, evaluation dimensions, safety gates, burden questions, and working proof artifacts.
Being tested
Transferability and validation
How the method performs across people, records, systems, professionals, settings, and different levels of human capacity.
Not claimed
Clinical effectiveness or authority
No diagnosis, treatment recommendation, standard-of-care determination, clinician replacement, or validation of personal medical hypotheses.
08 · Contact
Continue the conversation.
Interested in the framework, longitudinal-AI evaluation, research, or a possible collaboration? Send Justin a private note.
A conversation, not a clinical service.Patient-Side does not provide diagnosis, treatment guidance, urgent support, or review of personal medical records through this form.
PS
09 · The work
Built from lived need. Developed with an analyst’s discipline.
Patient-Side began with a practical translation problem: consequential experience was being lost between life as lived and life as reconstructed. The public work extracts the generalizable method while keeping private clinical records private.