DijitalPi
TREN
Contact
Home/ AI Solutions
Occupational health and safety · Repetitive industrial workflows

AI System for Classifying Near-Miss Reports by Risk

Reports describing more than one hazard go directly to the health and safety specialist for review. The system proposes a risk class from the text, photograph, location and equipment; the specialist determines the official class and the next step.

Representative workplace safety ops panel with a work queue, checks and an audit trail: near-miss assessment queue
Near-miss assessment queue Representative interface — contains no real data. Worker identities are masked. The panel does not assign an official risk class or close an incident.

The problem

When incident text, photographs, locations and equipment records are held in separate sources, their fields must be matched by hand before a risk class draft can be prepared.

The health and safety specialist must compare every suggestion in the risk class draft with its source record; missing or conflicting evidence must not replace the final decision.

If an incident containing several hazards is not set aside for review, a serious hazard could be given a low priority; the record must not proceed without a human decision.

What we set out to improve

  • Time taken for a risk class draft and its supporting evidence to reach the health and safety specialist for review
  • How records set aside for human review because one incident contains several hazards are resolved
  • Number of suggestions corrected after identifying a risk that a serious hazard could be given a low priority

How the system works

  1. 01 Retrieve the incident text, photograph, location and equipment record from permitted sources using the record identifier
  2. 02 Link source, date and version information to the shared work record
  3. 03 Validate fields and units
  4. 04 Prepare the risk class draft with its supporting evidence
  5. 05 Check for an exception: one incident contains several hazards
  6. 06 Have the health and safety specialist accept, revise or reject the suggestion
  7. 07 Record the approved decision and its source in the processing log

Methods we used

  • A source-identified common data schema for incident text, photographs, locations and equipment records
  • A decision structure that separates deterministic rules from AI suggestions
  • A review screen showing the risk class draft alongside the underlying source evidence
  • A mandatory human review queue when one incident contains several hazards
  • An event identifier that prevents duplicate processing

Where people stay involved

The health and safety specialist compares the risk class draft with the supporting record. The workflow pauses when one incident contains several hazards. The responsible person can revise or reject the suggestion, or request further information.

Data and security

Only the fields required for this decision should be processed from incident text, photographs, locations and equipment records. Access to the source systems must be authorised separately from the health and safety specialist role, and personal data, commercial information and technical secrets must not be passed into the model context unnecessarily. Reading, suggestions and human decisions must be logged separately.

Who this suits

A good fit

  • Teams that regularly prepare risk class drafts
  • Organisations that combine incident text, photographs, locations and equipment records by hand
  • Companies that want people to retain control of exception decisions

Not a good fit

  • Operations where source data is not current and consistent
  • Operations where responsibility for the health and safety specialist role has not been defined
  • Operations where AI suggestions would be implemented without review

Frequently asked questions

What data does the system use?

Incident text, photographs, locations and equipment records can be received through an API or controlled file transfer. Fields whose source and version cannot be established are not treated as confirmed information.

When does the workflow stop?

The record goes to the health and safety specialist for review when one incident contains several hazards. If supporting evidence is missing, no action is taken in the target system.

Does AI implement the decision on its own?

No. The risk class remains a draft until the health and safety specialist approves it. Acceptance, revision and rejection are recorded in the processing log as human decisions.

Project led by:DijitalPi

✦ PI ASSISTANT · ARTIFICIAL INTELLIGENCE

online

Ask the DijitalPi AI Assistant your question

No need to fill in a form and wait for a reply — write your question and let Pi, trained on DijitalPi's 20 years of know-how, answer within seconds.

  • Instant answers, 24/7
  • Service-specific context — no generic replies
  • Can connect you to a free consultation if you wish
Hi! 👋 I'm Pi — DijitalPi's AI assistant. Write whatever you'd like to know about AI System for Classifying Near-Miss Reports by Risk and I'll answer right away.

Pi Assistant answers are for information purposes; for a proposal get in touch.

ASK AI ABOUT DIJITALPI

Let an AI explain what DijitalPi does.

Opens your chosen assistant with a ready research prompt. It reads the site live and answers.