Science

Methodology

How the Move Index works, what data it uses, and where the numbers come from. Full transparency into every assumption.

Overview

What Move Index Is

Move Index is a research-backed scoring system that helps desk workers understand the estimated value of short movement breaks. It produces a composite 0–100 score across four dimensions, always accompanied by confidence levels and caveats.

It is designed to compare short movement breaks within a desk-work context, not to rank all exercise universally. It is not a calorie calculator, a medical device, or a replacement for professional exercise prescriptions.

Plain English

The simple version

Move Index asks four simple questions before it scores a movement:

Question 1

How much of your body does this use?

Question 2

How much energy does it cost?

Question 3

How much does it undo desk posture?

Question 4

How realistic is it during a workday?

Framework

The Four Dimensions

MD

Muscular Demand

30% weight

Total muscle mass recruited multiplied by activation intensity. Weighted by relative muscle group size (gluteals and quadriceps contribute more than forearms).

Source: Published EMG studies (%MVIC)

MC

Metabolic Cost

25% weight

Energy expenditure above resting, scaled relative to brisk walking (3.5 METs) using square-root normalization for diminishing returns at high intensities.

Source: 2024 Compendium of Physical Activities

MV

Mobility Value

20% weight

Joint range of motion and posture correction potential. This is the most modeled dimension and should be interpreted as a directional estimate rather than a direct measurement.

Source: Biomechanics literature (mostly modeled)

DP

Desk Practicality

25% weight

Can you do this at your desk? Considers space, noise, equipment, recovery time, and professional appearance. Weighted highly because Move Index is designed for real-world work breaks, not lab-optimal exercise ranking.

Source: Expert heuristic (modeled)

Transparency

Data Classification

Every data point is classified by how it was obtained:

Measured

Directly from published research. Highest confidence.

Derived

Calculated from measured inputs via established formulas.

Modeled

Estimated from heuristics or analogy. Lower confidence.

28

total exercises

9

with measured EMG data

19

primarily modeled

Confidence

Confidence Scores

Each score includes a confidence level (0–1) computed from data coverage, study quality, population match, and data source type. The system never suppresses low-confidence scores — it shows them with appropriate caveats.

High (≥ 0.75)Strong research base with multiple corroborating studies
Moderate (0.50–0.74)Partial data or population mismatch
Low (< 0.50)Primarily modeled — interpret as directional
Development

How the Scoring Engine Was Built

The scoring algorithm was developed using an adaptation of Andrej Karpathy's autoresearch pattern. An AI agent iteratively modified the scoring logic, and each iteration was evaluated against a curated reference dataset using five metrics: score ordering, confidence calibration, consistency, interpretability, and simplicity.

Improvements were kept; regressions were reverted. The improvement history is itself part of the product — full transparency into how the algorithm evolved.

Evaluation

Evaluation Framework

The scoring engine is evaluated against five metrics (lower is better):

25%
Score Ordering

Do higher-value movements score above lower-value ones?

25%
Confidence Calibration

Do confidence bands match actual uncertainty?

25%
Consistency

Are similar exercises scored similarly?

15%
Interpretability

Are explanations clear, accurate, and appropriately caveated?

10%
Simplicity

Is the algorithm as simple as possible without sacrificing quality?

Limitations
  • Move Index scores are estimates, not clinical measurements. They should not be used for medical decisions.
  • Mobility Value and Desk Practicality are primarily modeled from expert heuristics, not direct measurement.
  • EMG data comes from studies with varying populations, sample sizes, and protocols. Individual results will differ.
  • Metabolic cost varies significantly with body weight, fitness level, and exercise form.
  • The system does not account for individual health conditions, injuries, or contraindications.
References

8 peer-reviewed publications underpin the Move Index scoring system. Access full text, PDFs, and formatted citations below.

1

2024 Adult Compendium of Physical Activities: A Third Update of the Activity Codes and MET Values

Barbara E. Ainsworth, William L. Haskell, ... et al.

Medicine & Science in Sports & Exercise(2024)·DOI: 10.1249/MSS.0000000000003297

Primary source for MET values across all exercises

ToolsDOIFull Text via DOIPublisher
2

Electromyographic activity in the gluteus medius, gluteus maximus, biceps femoris, vastus lateralis, vastus medialis and rectus femoris during the Monopodal Squat, Forward Lunge and Lateral Step-Up exercises

José M. Muyor, Isabel Martín-Fuentes, ... José A. Antequera-Vique

PLOS ONE(2020)·Vol. 15(4)·pp. e0230841·DOI: 10.1371/journal.pone.0230841

EMG activation data for squats, lunges, step-ups, and deadlifts

3

Key Factors Associated with Adherence to Physical Exercise in Patients with Chronic Diseases and Older Adults: An Umbrella Review

Daniel Collado-Mateo, Ana Myriam Lavín-Pérez, ... et al.

International Journal of Environmental Research and Public Health(2021)·Vol. 18(4)·pp. 2023·DOI: 10.3390/ijerph18042023

Rectus femoris and erector spinae activation during sit-to-stand

4

Sitting for Too Long, Moving Too Little: Regular Muscle Contractions Can Reduce Muscle Stiffness During Prolonged Periods of Chair-Sitting

Ewan Thomas, Guglielmo Pillitteri, ... et al.

Frontiers in Sports and Active Living(2021)

Evidence that brief periodic contractions prevent sitting-induced stiffness

ToolsPublisher
5

Comparison of energy metabolism and muscular activity between sitting on a stability ball and office chairs

Annals of Rehabilitation Medicine(2015)

Stability ball increases O2 consumption 13% and muscle activity 55% vs standard chair

ToolsPublisher
6

2011 Compendium of Physical Activities

Barbara E. Ainsworth, William L. Haskell, ... et al.

Medicine & Science in Sports & Exercise(2011)·Vol. 43(8)·pp. 1575–1581·DOI: 10.1249/MSS.0b013e31821ece12

Established MET reference values for common activities

ToolsDOIFull Text via DOIPublisher
7

Estimating energy expenditure during bodyweight resistance exercise

Journal of Sports Sciences(2019)·DOI: 10.1080/02640414.2019.1633030

MET values specific to bodyweight resistance exercises at various intensities

ToolsDOIFull Text via DOIPublisher
8

Non-exercise activity thermogenesis in the workplace

Reviews in Endocrine & Metabolic Disorders(2022)

Framework for valuing movement breaks in sedentary work contexts

ToolsPublisher