R3CAP Ops

Project Kanban

Every open non-EPIC issue and pull request in the R3CAP research, development, and CHI paper repositories becomes a card; closing or merging it removes the card at the next sync. Open pull requests appear in To Review, assigned issues appear In Progress, and unassigned or backlog issues appear in Todo. EPIC, repository, owner, contributor, and activity metadata drive the filters.

Proposal marks the latest detected suggestion awaiting a response from an assignee. ACK means an assignee replied or reacted; it records acknowledgement, not approval. Current week marks recent activity.

Details are available to authorised repository collaborators.

Todo 32

In Progress 5

To Review 6

Weekly reviewOutstanding PRs are reviewed here.Need team review? Open a PR before the meeting.

Team Practices

This working agreement makes feature ownership, proposed changes, and integration visible. It is intended to facilitate timely communication without hindering progress, discussion, or debate.

Owned feature
<owner>-branch
Team-reviewed PR
dev · shared merge authority
Suggested amendment
<suggester>-branch
Team-reviewed PR
<owner>-branch
Release integration
dev
Meeting decision, or as needed
main · shared merge authority
Issues / Ownership

One feature, one owner

  • Clean up the issue list so that each issue describes a single feature.
  • Each feature issue has exactly one assignee; that assignee is the feature owner.
  • The owner implements the feature in their owner branch and is responsible for bringing it to completion through a PR into dev.
Changes / Collaboration

Suggestions remain welcome

  • Any team member may suggest a change by replying in the feature's issue thread.
  • A suggester may also state that they will implement the proposed amendment in their suggester branch.
  • When complete, the suggester opens a PR into the owner's branch so the change remains visible and coordinated with the feature owner.
Review / Integration

Review branch changes together

  • For the current trial, the team reviews every PR in the web R3CAP meeting, whether it targets an owner branch or dev.
  • By default, the team also decides in these meetings whether to open a PR from dev to main; the team may do so outside the meeting when needed.
  • After the agreed review, any feature owner may merge an approved PR into dev or main; these integration branches are shared across the development team.
  • The review meeting includes a retrospective on whether these practices are helping and what should be adjusted.
Standup
Monday + Thursday cadence

Every team member posts a short standup update on both days. This twice-weekly cadence provides shared awareness while keeping the administrative load manageable.

Commercialisation Approach Under Exploration

The current working direction is to explore incorporating Immersification as a lean founder-led company that would house R3CAP and provide a home for future products. It would be a possible route for product validation and, where authorised, commercialisation. This is an exploratory concept rather than a final incorporation, ownership, or licensing decision.

Initial model

Immersification houses the portfolio

  • Immersification is envisaged as the company housing R3CAP and future products developed through the team's research and innovation work.
  • The founder-directors would initially run product validation, delivery, partnerships, and company decisions themselves.
  • No CEO, COO, or salaried operational employees are assumed at the outset.
  • Temporary responsibilities, signing authority, budgets, and decisions requiring whole-board approval would be documented clearly.
Lean support

Outsource statutory functions

  • Use an outsourced corporate secretary and external bookkeeping, tax, legal, IP, and data-protection support as needed.
  • Keep fixed staffing commitments low while the team tests demand, delivery obligations, and the viable commercial model.
  • Recruit a CEO or other executives later only when commercial activity and clear executive ownership justify those roles.
Decision gates

Resolve authority before exploitation

  • Seek and incorporate advice from relevant SIT colleagues and departments on IP ownership, licensing, conflicts of interest, and the appropriate institutional relationship.
  • Clarify rights and obligations arising from grant funding, project agreements, and collaborators.
  • Assess data-protection, healthcare, and regulatory obligations before making deployment commitments.
Current status: approach under exploration. Incorporating Immersification would not by itself establish ownership of R3CAP or permission to commercialise it. The proposed company and product-portfolio structure should remain provisional until the relevant SIT, IP, licensing, grant, collaborator, and conflict-of-interest questions are resolved.

Paper Collaboration

A paper repository contains the whole submission, just as a development repository contains the whole system. Ownership stays at the feature level: each paper section, analysis task, figure, video, representative image, or other submission artifact has one assigned owner. The first author stewards whole-paper integration and is the only person who merges into main; this does not make the first author the owner of every paper feature.

Owned paper feature
<owner>-branch
Owner prepares feature PR
PR to main
Suggested amendment
<suggester>-branch
PR to <owner>-branch
Feature owner incorporates
Paper integration
Feature-owner PR to main
First-author integration review
First author merges
Issues / Ownership

One paper feature, one owner

  • Use bounded issues for paper sections and submission artifacts such as analyses, figures, videos, representative images, and supplementary material.
  • Each issue has exactly one assignee; that assignee owns the paper feature from scoped work through a PR to main.
  • The repository and complete submission are the shared system-level artifact, not a single ownership unit. The first author may own assigned features like any other contributor.
Changes / Collaboration

Suggestions stay with the feature owner

  • Reply in the relevant issue or team thread with the specific change you suggest.
  • State whether you are proposing an idea or offering to implement it, and check what the feature owner and other contributors are already doing.
  • For an implementation, work in your suggester branch and open a PR into the feature owner's branch. The feature owner remains responsible for the eventual PR to main.
Review / Integration

First author integrates main

  • The feature owner reviews and consolidates work within their owned scope; paper PRs do not require review by the whole team or a scheduled meeting.
  • The first author reviews PRs to main for argument, structure, cross-feature coherence, and submission readiness, and is the only person who merges them.
  • Anyone may build locally. Canonical submission builds come from main under first-author coordination.
Same feature ownership, different merge gate. Development integration branches have shared merge authority among feature owners; paper main has first-author-only merge authority. Announce substantial work and coordinate first with the relevant feature owner.

Research Questions

The current CHI framing centres healthcare professional (HCP) experience while also exploring the technology's educational prospects. The updated RQs below address how HCPs experience, compare, and evaluate R3CAP for remote home assessment, and how its real-time 3D reconstruction capabilities might support education and training.

RQ1 / Design

Overarching HCP experience

How can 3D reconstruction technology be designed to support HCPs in performing remote home assessments?

  • Identify which unique features R3CAP offers to HCPs' experiences, including real-time progressive reconstruction, AR-enabled spatial recommendations, spatial notes, visual accuracy, spatial avatars, and related capabilities.
  • Use concurrent think-aloud during use and post-session semi-structured interviews as the main data sources.
  • Apply semi-deductive thematic analysis with predefined component-level codes to surface design nuances, breakdowns, and opportunities.
RQ2 / Workflow

R3CAP versus current practice

How does the experience of using R3CAP compare with traditional HCP workflows for home assessment?

  • Compare R3CAP against recalled traditional workflows using interview prompts and workload/usability items that distinguish retrospective baseline assessment from live R3CAP use.
  • Use NASA-TLX immediately after R3CAP conditions; if used for traditional workflow comparison, frame it as a retrospective workload estimate rather than an equivalent live measurement.
  • Treat post-session semi-structured interviews as the primary evidence, with workload ratings and descriptive statistics as supporting evidence.
  • Analyse both quantitative workload/usability patterns and thematic accounts of what R3CAP changes in assessment practice.
RQ3 / Education

Educational prospects

How might the real-time 3D reconstruction capabilities enabled by R3CAP support education and training?

  • Explore prospective teaching and training applications, relevant learners and educators, useful learning scenarios, and perceived advantages or limitations.
  • Address this exploratory RQ through additional questions in the post-session semi-structured interview only; it does not add another task, condition, or instrument.
  • Analyse responses thematically as prospective opportunities and requirements, not as evidence that R3CAP already improves educational outcomes.
RQ4 / Interface

DESKTOP versus VR

How do DESKTOP and VR interfaces compare in usability, workload, presence, simulator discomfort, and perceived value for HCPs experiencing 3D reconstructed environments during remote home assessment?

  • Use VRSQ to capture VR-related symptoms and discomfort where relevant.
  • Use IPQ to compare spatial presence and immersion between interface conditions, noting limits when applying presence measures to DESKTOP versus VR.
  • Use NASA-TLX and usability ratings as secondary comparative measures across DESKTOP and VR.
  • Use post-session interviews as the main source for interpreting why HCPs prefer, trust, struggle with, or value each interface.
  • Analyse interface differences through descriptive statistics and thematic analysis rather than over-claiming inferential effects from a likely small CHI study sample.

MedicaTwin Session Procedure

The session is a single ~95 minute within-subjects visit. Each session pairs one HCP with one scanning collaborator. Condition order is counterbalanced: roughly half DESKTOP-first and half VR-first. Between conditions, the simulated home is rearranged or a different room is used to reduce memory effects.

Active recording set for each condition: think-aloud audio, SC device screen, HCP editing-client screen, video of HCP, and video of SC.
  1. Recruit, screen, and schedule. Up to 20 HCPs and up to 20 SCs, aged 21-65, are recruited through professional, clinical, convenience, and snowball routes. HCP screening captures role, institution, years of clinical experience, home-assessment frequency, practice areas, HomeSNAP or similar-tool experience, TTSH connection, 3D/digital-twin familiarity, and VR familiarity. SC screening captures occupation, study relationship, HomeSNAP or similar-tool experience, TTSH connection, 3D/digital-twin familiarity, and VR familiarity.
    ParticipantsOne HCP and one SC per session.
    TokenS$30 voucher per participant.
    SiteSIT, TTSH, or a suitable site with an HCP workstation and simulated home environment.
  2. Orientation, consent reconfirmation, and familiarisation (~15 min). The facilitator briefs the HCP on the session procedure, walks them through the r3cap interactive tutorial, explains editing-client navigation, tools, and what they are seeing, then gives a few minutes of free access and invites questions. The SC is briefed on their role, familiarises with the physical environment layout, notes objects in the room(s) and what they may want to scan, learns capture-client buttons and view state, receives a few minutes of free access, and can ask questions. Consent is reconfirmed for both participants, including explicit understanding of recording.
  3. Condition 1 remote assessment (~15 min). The first modality is assigned by the counterbalanced order: DESKTOP or VR. The HCP and SC connect over a live video feed, introduce themselves, and receive the mock patient's profile from the facilitator. The SC begins live capture of the first simulated home environment. The HCP views the real-time reconstruction in the editing client, thinks aloud, observes the SC's scanning actions, and performs assessment tasks. Tasks are intentionally not fixed in a strict order; breadth of experience and rich qualitative data matter more than complete task throughput, while the facilitator still ensures the participant works through the task set at their own pace.
    HCP viewDESKTOP display with keyboard and mouse, or VR headset with controllers.
    SC viewLiDAR-equipped mobile capture client in the simulated environment.
    RecordingsThink-aloud audio, both screens, HCP video, SC video.
  4. Transition (~5 min). The facilitator rearranges the simulated home environment or moves the session to an alternate room, switches the HCP to the other viewing modality, resets and re-arms every recording channel for Condition 2, and offers a short break. The transition is part of the design because the second condition should present a meaningfully different spatial layout rather than only repeat a memorized room.
  5. Condition 2 remote assessment (~15 min). The HCP repeats the same assessment-task family using the other interface, while the SC captures the new or rearranged space. The same recordings begin again. This second condition is the within-subject comparison point for usability, workload, collaboration quality, VR symptoms, task metrics, and later interview reflections.
  6. Post-session surveys (~15 min total). Surveys are administered electronically via Qualtrics once per condition, about 5-7 minutes each, with adequate breaks between conditions. HCP surveys include SUS, NASA-TLX, collaboration quality, VRSQ after VR only, comparison to conventional methods if relevant, and overall impressions. SC surveys include SUS, NASA-TLX, collaboration quality, and overall impressions.
  7. Semi-structured interview (~30 min). The HCP and SC are interviewed together. The guide covers reflections on both conditions, collaboration experience, perceived value versus current practice, preference between DESKTOP and VR, suggestions for improvement, and the educational prospects of R3CAP's real-time 3D reconstruction capabilities. The additional education questions explore possible teaching or training uses, intended learners and educators, useful scenarios, benefits, limitations, and requirements. Interview prompts are aligned to RQ1-RQ4; RQ3 adds interview questions only and does not add another study task or condition. The repository marks the standalone interview guide as a remaining TODO.
  8. After-session closeout. Stop and save all recordings; verify files are captured and named with de-identified IDs; confirm both surveys were submitted; capture session metrics; issue the S$30 voucher to each participant; finalize facilitator notes; and reset the room or environment for the next session.

Assessment Task Detail

The condition tasks are deliberately flexible in order, but detailed in content. The facilitator should make sure participants encounter enough of this task family for rich qualitative reflection.

HCP task family
  • Observe the SC as they scan and reconstruct the environment.
  • Guide the SC on missed or important areas, such as a missed hallway or partially reconstructed doorway.
  • Communicate potential patient issues as scanning continues.
  • Identify, annotate, and label hazards such as loose rugs or narrow doorways.
  • Measure dimensions such as doorway width or bed height.
  • Document issues such as a bedroom door being too narrow.
  • Communicate and document recommendations such as removing a rug or replacing it with a non-slip mat.
  • Use annotations, measurements, markings, and virtual modifications, including adding objects such as grab bars or mobility aids.
  • Perform a first-person walkthrough to simulate patient navigation, including walking-aid or wheelchair navigation.
SC task family
  • Familiarise with the simulated home layout before or during orientation.
  • Operate the LiDAR-equipped mobile capture device and live-scan the physical environment.
  • Communicate with the HCP in real time while scanning.
  • Relay issues the mock patient is assumed to face.
  • Respond to HCP requests for closer views or specific areas.
  • Adjust scanning focus as the HCP identifies uncertainty, missed geometry, or clinical concerns.
  • Collaborate through the platform while preserving the remote-assessment scenario rather than acting as an expert assessor.

Data Collection and Instruments

Qualitative materials are primary for CHI claims. Survey responses add descriptive support, while the recorded Zoom materials preserve the think-aloud, observation, and interview evidence needed for thematic analysis.

Qualtrics

Surveys

  • Administer post-condition surveys electronically through Qualtrics.
  • HCP instruments include SUS, NASA-TLX, collaboration quality items, VRSQ after VR only, current-practice comparison items, and open comments.
  • SC instruments include SUS, NASA-TLX, collaboration quality items adapted for scanning, overall impressions, and open comments.
  • Use survey results as descriptive support for RQ2 and RQ3 rather than as the sole basis for CHI claims.
Zoom recording

Think-aloud

  • Record HCP think-aloud audio during each DESKTOP and VR condition through Zoom.
  • Prompt HCPs to verbalize uncertainty, perceived value, spatial understanding, trust, and breakdowns while they use R3CAP.
  • Use the transcript as a primary qualitative source for component-level design nuances in RQ1.
Zoom recording

Observations

  • Record HCP screen, HCP video, SC capture-client screen, and SC video through Zoom where feasible.
  • Capture gestures, navigation behaviour, visible confusion, requests for rescans, annotation and measurement activity, and SC response to HCP guidance.
  • Extract session metrics such as time on task, hazards identified, annotations, recommendations, and measurements as supplementary evidence.
Zoom recording

Semi-structured interview

  • Record the end-of-session semi-structured interview in Zoom.
  • Interview HCP and SC together, with prompts focused on both modalities, perceived clinical value, comparison with current practice, interface preference, improvement suggestions, and prospective education or training uses.
  • Map interview evidence back to RQ1-RQ4, treating RQ3 as an exploratory thematic account of educational prospects rather than an evaluation of learning outcomes.
Analysis

Planned interpretation

  • Semi-deductive thematic analysis of think-aloud, observation, interview, and facilitator-note materials.
  • Descriptive survey summaries with means, standard deviations, and distributions.
  • Optional paired non-parametric condition comparisons where appropriate.
  • Session metrics compared between conditions as supplementary evidence.

Team Retrospective — Discussion Points

01

Lead and process gap

  • Discuss the project lead's role in improving the process.
  • Confirm plan ownership is visible and accurate.
02

Shared care, neutral frame

  • Quality belongs to the whole team.
  • Discuss process and working-style alignment, not personal judgement.
03

LISTEN FIRST

  • Ask everyone what helps.
  • Ask everyone what makes work harder.
  • Listen before proposing a solution.
04

Reflect shared needs

  • Summarize what the team heard.
  • Confirm shared understanding.
05

Keep / Fix / Drop

  • Discuss what helps, blocks, or adds noise.
  • Agree on a two-week process trial together.
06

Ownership through ready-for-integration

  • Agree who drives each issue end to end.
  • Define clear scope and acceptance for handoff.
07

Integration versus redesign

  • Clarify when work is integration.
  • Treat substantive redesign as a new team decision.
08

Visibility, reviews, decisions

  • Choose a lightweight standup or shared view.
  • Agree on PR review before merge.
  • Record decisions, owners, and next steps.
09

Missing loop: agree → try → check back → improve

  • A one-off solution is not the finish.
  • Lead facilitates regular review with the team.
  • Review the trial together and adjust.

Updated for CHI planning. Visual template follows the SplatBench planning page color system and procedure-card flow.