Roster history
Published, revised and actual duties. Source and timing retained.
MAFRM / 01
The platform
One connected view.
Clearly separated evidence.
MAFRM connects roster history, continuous fatigue modelling and independent observations while keeping each evidence type clear and traceable.
System architecture
From operational records
to explainable questions.
The architecture links source records, model assumptions, scenario inputs and independent observations into one inspectable analytical workflow.
Rules & assumptions
Versioned requirements, model configuration and data-quality checks.
Scenario inputs
Isolated alternatives, each compared with the same baseline.
Continuous, explainable analysisSeparate roster checking and fatigue modelling · Reproducible run identity
Review & compare
Duty-level explanations, downstream effects and population patterns.
Independent observations
Sleep diaries, wearables, fatigue ratings and functional tests.
Evaluate & improve
Validation layer with frozen predictions kept separate.
Current capability
Begin with a trustworthy roster.
Duty records are the starting point. Their meaning depends on the source, activity codes, time zones and whether they represent planned or actual work. Missing or ambiguous data should remain visible.
Selected workbook and JSON imports are supported in the current prototype. The roadmap adds a broader event schema, authorised provider adapters and a full published–revised–actual history.
Current capability
Keep rules and predictions distinct.
A roster checker asks whether recorded facts satisfy a configured rule. A fatigue model asks what those inputs imply under its assumptions. Keeping them separate makes both results easier to understand.
MAFRM maintains a separate roster checker and continuous calculation. Generalised operator rule packs, effective dates and fuller calculation traces are in development.
Development foundation
Follow state across time.
A continuous model retains the effect of preceding duties and recovery opportunities across roster boundaries. Reviewing a new month should not erase the history used to establish the starting state.
Reproducible analysis also requires the input, model version, variables and assumptions to travel with each result. Content-addressed run identities provide the foundation for durable records and broader calculation traces.
Scenario capability
Compare a change with its baseline.
The Scenario Lab can isolate a single-duty edit and retain named A/B/C scenarios in browser storage, including warnings when options do not share a common baseline.
The roadmap extends this capability into added, removed and resequenced events, controlled sensitivity analysis and hosted scenario records.
Validation program
Bring observations alongside predictions.
The observation layer is designed to hold sleep diaries, subjective ratings, suitable wearable observations and functional tests separately from roster-derived predictions.
The validation workflow freezes a prediction, collects observations independently, aligns the records, assesses agreement and investigates errors.
Individual to organisation
Detail when you need it.
Context when you step back.
Individual analysis can help explain a particular duty or transition. Population views can help identify recurring patterns worth investigating across a larger roster.
MAFRM includes population review views. Organisational reporting is being designed around identity, access controls, privacy thresholds and tested multi-organisation separation.
Explore data & integrationsHelp shape what comes next
A better question starts
with a conversation.
Developing a roster proposal, investigating an operational pattern or planning fatigue research? Let’s explore where MAFRM could contribute.