Detecting anomalies

Threats and anomalies are a challenge that public organizations face in many ways in their daily mandate for the citizen. Whether it be for doctors to identify health anomalies, for city authorities to detect broken infrastructure, or tax authorities to tackle the issue of fraud, public servants have to act quickly and efficiently. Sometimes, it also means staying one step ahead of criminal behavior.

Why is it important?

Sixty-one percent of companies stated in a recent study[1] that AI is vital to identifying critical threats. This also applies to the public sector, where various anomalies (structural or punctual, directly visible or otherwise) must be identified in real time in order to cope with potential threats. The challenge for the public sector is to identify a detection mechanism of anomalies that works in real time and with the best possible precision while conforming to jurisdictional requirements. That said, there is no effective way to tackle anomalies using a thumb rule. The need to constantly analyze the growing data manually is next to impossible. As everything is constantly evolving, “normal” is also continuously being redefined. In this context, leveraging AI can bring the analytic augmentation a public organization needs to identify anomalous behavior or patterns in real time.

AI for the rescue

AI enables the identification of anomalies using various types of data that can be numerical, image, audio, or text based. Building on the existing data pool, AI can detect patterns suggesting immediate cases of danger or unlawfulness. Furthermore, AI can detect all kinds of anomaly stages whether they are suspicions, an identified threat, or an incident occurring. The more data available, the better the quality of the AI in detecting anomalies or situations of danger.

Capgemini accompanies the full scope of public organizations by extending their preparedness for dangerous situations with AI & Data Analytics – from public administrations up to healthcare professionals (see table).

Graph 1: Use Cases for “Anomaly Detection” in the Public Sector