Applied Research for Complex Environments
The research program of LeanLogic Systems, LLC
Pre-registered, reproducible studies of public enforcement records, beginning with whether documented high-energy hazards predict the serious injuries that follow.
The public record contains predictive information it does not expose. Our work is to recover it, test it against what actually happened, and report the result, including when the answer is no.
The Question
Every year OSHA documents hundreds of thousands of workplace hazards: a missing guardrail at height, an unshored trench, exposed conductors, an unguarded machine. Each is recorded, cited and, usually, abated. The record then goes quiet.
Serious injury follows an uncontrolled release of energy: gravitational, mechanical, electrical. The hazards OSHA documents are, in physical terms, energy that has not yet been released. So the question is direct: when an employer has been documented with an uncontrolled high-energy hazard, is a serious injury or death from that same energy more likely to follow, beyond what the employer's size, industry and prior injuries already predict?
If the answer is yes, the record already holds a signal that could direct scarce preventive attention. If no, that too is worth knowing. Either way, the answer has to be earned under a design written down before anyone looked.
The Study: Energy-Specific Recurrence
The lead study tests whether documented uncontrolled high-energy exposure at an employer (identified through a prior OSHA citation referencing a standard that establishes an energy threshold, or through a prior serious injury or fatality) predicts subsequent serious injury or fatality involving the same energy class, beyond the predictive contribution of prior injury history, establishment size and industry.
| Registration | Open Science Framework, t6gds, public since 15 October 2026 |
|---|---|
| Design | Retrospective employer-level cohort on national OSHA inspection, citation, accident-investigation and Severe Injury Report records, 2005–2026; five energy families (fall, struck-by, caught-in, electrical, trench); placebo citations and negative controls; out-of-time validation |
| Data | Public records only, acquired as a single-day snapshot whose hashes are registered; no worker or victim identifiers enter any analysis |
| Status | Registered analyses complete; manuscript in preparation for the American Journal of Industrial Medicine |
| Investigator | Michael Beauchamp, ORCID 0009-0009-6039-3970 |
Secondary and follow-on questions are registered or under development alongside it: whether the inspector's own assessment of a hazard's gravity carries additional predictive information; whether the type of control an employer installs, engineered or administrative, changes what follows; and whether the same relationships hold across industries.
Method
Every study in the program is designed before it is executed. Hypotheses, classification rules, test scripts, data snapshots and pass/fail criteria are written, hashed and publicly registered first. Results are then reported against the registration, whether supported, inconclusive or withdrawn.
Pre-registration. The protocol, its rules and its tests are frozen and hash-chained; any change after registration is a logged deviation, never a silent edit.
Reproducibility. Analyses run in a pinned computational environment on a registered data snapshot; a rerun reproduces the registered output byte for byte.
Symmetry. Negative and null results are reported with the same prominence as positive ones. A hypothesis that fails its test is withdrawn in print, not quietly dropped.
Boundaries. The program measures and reports. It does not adjudicate, and it does not identify individuals. Compliance and enforcement determinations remain with the institutions authorized to make them.
No result is characterized publicly before its record is complete, and when results are stated they are stated with their limitations attached.
Program
Paper 1. Energy-Specific Recurrence on the national enforcement record. In preparation; target journal the American Journal of Industrial Medicine.
Funded extension. A NIOSH exploratory research proposal in preparation, asking what further predictive information the public record holds but does not expose: how much energy was present, what the inspector judged, and what control was installed.
Beyond the record. Whether explicit hazardous-energy characterization improves AI-assisted visual hazard assessment compared with visual recognition alone. A later horizon, dependent on partners and appropriately governed data.
Methods validated through the program may be licensed to established safety-technology providers. Publication comes first; licensing follows what has been shown, and never precedes it.
Collaborate
The program develops research relationships selectively, around questions the registered work has opened.
Occupational epidemiologists and biostatisticians: independent methodological review of the registered design and its follow-on studies.
Energy-based safety researchers: extending established constructs of hazard magnitude and direct controls onto the national enforcement record.
Safety-technology providers: validation partnerships that test, rather than assume, whether hazard scoring and control-gap prioritization predict subsequent serious injury.
Research sponsors and institutions: the questions sit at the intersection of injury surveillance, enforcement data and prevention science.
Start a conversationAbout LeanLogic Systems
LeanLogic Systems, LLC is a Tennessee limited liability company based in Knoxville. It conducts applied research under the Exauren name and is the developer and operator of the program described here.
The research is led by Michael Beauchamp, founder and sole investigator. His background is in enterprise data architecture and applied statistical modeling in regulated, high-volume operational environments, including more than a decade building analytics and data infrastructure inside a Fortune 500 supply-chain organization, and earlier data engineering on healthcare claims and Medicaid member data. He holds an M.S. in Predictive Analytics from Northwestern University, a B.S. in Mathematics, and a certificate in Applied Data Science from MIT Professional Education.
Earlier Exauren work on deterministic evaluation of regulatory obligations is preserved as lineage. The program's focus is now occupational-safety research and the dissemination of what it validates.
The Mark
The spiral represents successive examination: repeated passes through incomplete evidence, each capable of improving the next and converging toward a better-supported account of reality.
The crosshairs establish a disciplined reference frame. Attention is directed by declared conditions and evidence, not merely by what is most visible.
The gold datum represents the condition under examination: the reality toward which the evidence must converge but which the program is never entitled to presume.