Decision Intelligence Guardrails (DIG) Example:

Four practical examples showing how DIG improves judgment across health, performance, and risk evaluation.

Confirmed COVID (Medical Attribution)

Read more

Excess Mortality (Statistical Approach)

Read more

Narrow Biological Attribution

Read more

Top-1% Bench Press at Age 65 (General Population)

Read more

Identify the Claim

What is the probability and severity of the adverse event if no action is taken?

Claim:

"X disaster caused Y deaths."

DIG immediately asks:

What time horizon?

What causal definition?

What evidence standard?

Without those, the claim is ambiguous.

Three Different Valid Answer Sets

1. Confirmed Immediate Deaths (Highest Evidence)

Definition:

Deaths directly attributable and medically documented within the event window.

Example:

Event	Immediate Deaths
Chernobyl	~30
Fukushima	0

Evidence standard: high medical certainty

2. Epidemiological Long-Term Radiation Effects

Definition:

Projected cancer deaths statistically attributable to radiation exposure over decades.

Example:

Event	Estimated long-term deaths
Chernobyl	~4000 (WHO/UNSCEAR estimate among most exposed populations)
Fukushima	Very small projected increase

Evidence standard: population-level statistical modelling

These deaths cannot be individually identified

3. Total disaster related mortality

Definition:

All deaths caused by social and indirect consequences.

Example:

Event	Disaster related deaths
Chernobyl	Thousands were included relocation health system disruption
Fukushima	~1,000-2,000 (mostly evacuation-related)

Evidence standard: broader public-health attribution

Why DIG Flags the Question

Dig would classify the claim:

"Deaths caused by Chernobyl Fukushima."

as:

Ambiguous Causal definition

Because it conflates:

Radiation deaths

Statistical cancer projections

Indirect disaster mortality

DIG Admissible Question Versions

DIG would require the question to be rewritten as one of the following:

"How many people died from acute radiation exposure within months of the accident?"

"What is the estimated number of long-term cancer death attributable to radiation exposure?"

"What is the total number of disaster related deaths including indirect effects?" Each produces a different correct answer.

Why This Example Is Valuable for DIG

It demonstrates four core DIG principles:

Definition locking

Terms like caused by must be defined.

Time horizon clarity

Immediate vs lifetime risk.

Evidence hierarchy

Confirmed deaths vs statistical estimates.

Policy implications

Different definitions drive different
policy conclusions.

Why It’s a Strong Teaching Example

DIG would require the question to be rewritten as one of the following:

Because the same event produces answers ranging from:

30 deaths → thousands → tens of thousands

depending entirely on definitions and modeling choices.

That’s exactly the kind of ambiguity DIG is designed to expose.

DIG Example: COVID Death Counts

1. Confirmed COVID Deaths (Medical Attribution)

Definition:

Deaths officially recorded by governments as caused by COVID infection.

Global total reported to the World Health Organization: ≈7 million deaths

Evidence standard:

Positive test

Physician attribution

Death certificate

Limitations:

Testing shortages early in the pandemic

Inconsistent reporting across countries

2. Excess Mortality (Statistical Approach)

Definition:

Deaths above expected baseline mortality during the pandemic.

Major estimates:

World Health Organization: ≈15 million excess deaths

The Economist: ≈20 million excess deaths

Evidence standard:

Statistical modeling of total mortality

Includes deaths from:

COVID infection

Healthcare system disruption

Delayed medical care

Indirect pandemic effects

3. Narrow Biological Attribution

Definition:

Deaths directly caused by viral damage to the body.

This number is unknown, because many deaths involved:

Pre-existing illness

Multi-factor causes

This category is not directly measurable.

DIG Example: Top-1% Bench Press at Age 65

1.Top-1% Bench Press at Age 65(General Population)

Definition:

A reported strength performance compared against age-adjusted population benchmarks.

Claimed Bench Press Performance 120 KG at age 65

Evidence standard:

Recorded Lift

Bodyweight Context

Standardised Range of motion

Limitations:

Rare but plausible with long-term training history

Percentile depends on body weight and lifting standard

Why DIG Flags the Claim

The statement:

"COVID killed X million people"

is not admissible without defining the measurement frame.

The answer depends on:

Whether deaths must be individually verified

Whether statistical attribution is allowed

Whether indirect effects are included

DIG Admissible Versions

DIG would require the claim to be rewritten as:

Confirmed deaths recorded by governments

Estimated excess mortality during the pandemic

Deaths directly caused by viral infection

Each produces a different legitimate answer.

Why This Example Is Powerful for DIG

It demonstrates three common data problems:

Definition ambiguity

Measurement limitations

Statistical inference vs direct evidence

The range:

7 million → 20+ million

is produced entirely by methodological choices, not disagreement about the underlying event.

Core DIG Lesson

Before evaluating a claim, the system must first determine:

What definition is being used

What evidence standard applies

What time horizon is included

Each produces a different legitimate answer.