Predictive Safety Analytics

Improve safety performance through speedy and informed decision making

Improving health and safety

Operational safety risk management is a challenge that has the potential to impact a company’s regulatory compliance, internal policy management, brand, reputation and finances.

By combining powerful statistical methods and leveraging multiple disparate data sources, organisations are able to understand the drivers of workplace accidents that were previously unseen. Companies can then use these insights to more effectively address workplace safety in terms of both injury prevention and injury management.

PwC has developed Predictive Safety Analytics, an approach which uses a range of sophisticated analytics techniques, to provide greater insight and clarity into health and safety systems, and to help its clients develop processes and design interventions to minimise such risk. For example, our approach can help you answer:

What metrics can provide the ability to take a proactive evidence-based safety focus on “leading indicators” to create more actionable insights, rather than the traditional simple reactive reporting of claims and incidents?
Which work processes or operational areas have the highest risk factors for incidents?
What would be the impact of proposed changes to training, teaming and rostering?
Given the highly variable nature of accidents, what preventative measures offer the best value across a range of future outcomes, when a safety program has budgetary constraints?

Predictive Safety Analytics

1. Business understanding

Understand project objectives and requirements from a business perspective.

This phase begins with agreeing the health and safety challenge(s), the approach for integrating the insights from the project into the business, defining the operating environment and determining the available assets and finally setting an “analytics plan” to achieve these outcomes.

2. Data understanding

Collect data and then become familiar with it (quality and initial insights).

The data extracts from the data assets are gathered. This typically includes: safety claims, incidents and observations, operational data, HR information, work site data, production data and other external data sources (e.g. geospatial socio demographics).

3. Data preparation

Construct modelling data by aggregating, manipulating and joining.

We integrate and manipulate your business data with our own external data sources to create an integrated data set that is ready for analysis.

4. Modelling

Choose and apply various analytics modelling techniques.

During this phase we apply powerful statistical techniques in order to discover and explain relevant relationships between safety outcomes and operational metrics. Combining traditional safety data with non-traditional sources (e.g. Census data) can lead to predictions about where accidents are most likely to happen, under what circumstances, and to which segments of the workforce—all before they actually happen.

5. Evaluation

Evaluate analysis in the context of the business issues being addressed.

The model results are validated using statistical validation techniques. This phase typically involves a series of interactive workshops with the business to explore and contextualise the analytical findings. Furthermore, the application of advanced cost optimisation modelling can help facilitate an objective assessment of the relative benefits of different safety spending options.

6. Deployment

Organise, present and deliver the insights in a way the business can use it.

The output from all previous phases is of little value if nothing is done with it. This phase covers the change management and best practices to build buy-in for predictive analytics to help bridge the gap between building analytical models and real world outcomes, including the monitoring requirements to drive proactive engagement across the organisation.

“The need to operate safely hasn’t changed – it continues to be fundamental to the management of a responsible business. It is the breadth and volume of data that's now generated as part of doing business which has - and that data is rarely used to inform safety decisions.”

Phil Bolton, Director