Modelling the Impact of AI-Driven screening vs. PCR Testing on Mpox Surveillance and Outbreak Control in Democratic Republic of Congo

Background

The Democratic Republic of Congo (DRC) faces significant challenges in controlling infectious disease outbreaks due to limited healthcare resources and infrastructure. This study models the potential impacts on disease spread and mortality rates, comparing high-accuracy traditional PCR testing versus a relatively lower-accuracy AI-driven rapid screening to detect and isolate approach during a hypothetical Mpox outbreak over a 12-month period.

Methods

We developed a computational model to simulate Mpox spread in DRC's population of 95 million, assuming a worst-case scenario of 30% infection without intervention. Two testing strategies were compared:

1. PCR Testing: 98% accuracy, 48-hour result time, capacity to test 100 individuals daily.

2. AI Tool: 94% accuracy, instant results, capacity to test 10,000 individuals daily.

Key Assumptions:

• Initial prevalence: 5%

• Basic reproduction number (R₀): 2.3

• Generation time: 12 days

• Mpox characteristics: Recovery period: 14 days Mortality rate: 3%

• Testing costs: AI tool is 99% cheaper than PCR per test

• Continuous daily testing for 365 days

• Detected cases are immediately isolated

• No additional public health measures beyond testing and isolation

Results

After 12 months of simulated outbreak:

Discussion

The model suggests that the AI-driven approach could significantly outperform traditional PCR testing in controlling a large-scale Mpox outbreak in DRC when resources are limited. Key findings include:

1. Infection Control: The AI tool potentially reduces total infections by 87.4% compared to PCR testing.

2. Mortality Reduction: Approximately 747,000 lives could be saved using the AI approach.

3. Long-term Impact: The AI tool shows potential to prevent nearly 25 million cases over 12 months.

4. Resource Efficiency: Given the assumed 99% cost reduction per test, the AI approach appears substantially more cost-effective.

Limitations

This model is based on simplified assumptions and does not account for factors such as geographical distribution, healthcare system capacity, or additional intervention strategies. Real-world application would require further validation and consideration of practical implementation challenges.

Conclusion

This modelling study suggests that an AI-driven rapid testing approach could dramatically improve Mpox outbreak control in resource-limited settings like DRC. The potential for saving lives and preventing infections underscores the need for further research into innovative, scalable testing strategies for infectious disease surveillance and control.