AI in Public Sector: Examples from Public Health & Government

AI adoption in the public sector is already noticeable in public health management, showing good examples of using AI with the aim to protect and improve people’s health, promote healthy lifestyles, prevent diseases, and do research.

In other words, it involves optimizing health services from the local to the government level, targeting healthcare’s biggest daily challenge: changing patient behavior. 

We know that nearly 70% of premature deaths are actually linked to modifiable behaviors such as smoking, sedentary lifestyles, poor diets, and failure to follow prescribed treatment plans, screenings, or medication regimens. This data, while saddening, points to the fact that behavior change is possible but only with the right strategy and implementation. 

The following case studies show how combining behavioral science with AI can most effectively improve public health outcomes.

1. Reducing patient no-shows in healthcare

Missed appointments cost the NHS around £1.2 billion annually in the UK. Despite common belief, forgetfulness is not the main reason for patient no-shows in clinics. 

Why else would a patient miss an appointment? An older patient might want to avoid rush hour or coordinate multiple appointments on the same day to reduce the number of trips needed. A parent may need to schedule around school drop-offs. How can AI technology be used to notice these circumstances?

Using AI to increase patient show-ups

Deep Medical accurately identified that there can be deeper causes that prevent patients from attending their appointments which require a better understanding of these patient groups and their conditions.

Their AI solution analyses various external factors, such as transportation availability, weather conditions, and individual patient circumstances like work schedules or caregiving responsibilities, to predict which patients are at risk of missing their appointments. 

By identifying these high-risk patients, the AI takes proactive steps to engage them early, offering appointment times that are more convenient or rescheduling in advance to reduce the likelihood of no-shows. 

2. Public Sector AI Adoption: Boosting Early Screening Participation

Early detection through public screening programs for diseases like cancer and diabetes significantly improves outcomes; for example, early-stage cancer diagnosis leads to a 5-year survival rate of 90%, compared to just 5-20% for late-stage detection. 

However, participation in UK screening programs remains unsatisfactory, with only about 65-70% of eligible individuals attending cancer screenings, leaving a significant portion of the population at risk for undiagnosed conditions and missed opportunities for early treatment.

Why is that? On one hand, many people feel healthy and don't see the need for screening, especially without severe symptoms, and many are unaware of available community-based screenings—in one study, over 58% of participants didn’t know about local screening services. 

Furthermore, emotional barriers like fear, practical challenges such as scheduling appointments, commuting to the clinic during working hours, and low motivation, often combine to result in inaction. So how can this be changed?

Using AI to identify high-risk patients and send tailored messages

Medial EarlySign uses AI-powered predictive models to analyze existing clinical data and identify individuals at higher risk of serious conditions like colorectal and lung cancer. Their algorithm detects subtle patterns in patient data that may go unnoticed by traditional methods. Once high-risk patients for early screening are identified, AI can nudge them to participate in screenings through personalized text messages. 

A US study using LLMs (large languaging models, a type of AI algorithm) could detect these emotional barriers based on words like “scared”, “worried”, and “anxious”, and respond with tailored messages to address patients’ concerns with a message like this:

“We understand that it can be stressful to wait for test results. But the results are so important because they tell you about your colon health. The good thing about catching problems early is that they can be treated.”

3. Helping diabetes prevention by managing glucose levels

Continuous Glucose Monitoring (CGM) devices enable patients to track their glucose levels in real-time and see how meals and activities affect them. 

These devices can significantly reduce hypoglycemic episodes (when a person’s blood sugar is too low), with mild cases dropping from 4.75% to 0.78% and severe cases from 3.01% to 0.2%. This reduction leads to fewer emergencies, fewer hospital visits, and lower healthcare costs.

Despite their benefits, CGMs are often limited by high costs and inconsistent insurance coverage, which hinders widespread use. In a study, more than 55% of surveyed patients cited cost as the main barrier to try CGMs. 

Removing barriers to manage glucose levels with AI

The January AI app simplifies glucose management by using predictive AI and image recognition to show the impact of food before it's eaten, without costly CGM sensors. 

Users can scan a barcode or take a photo of their meal to instantly identify ingredients and predict their effect on their glucose levels. Based on scans, the app offers personalized nutrition guidance and nudges users toward healthier alternatives right when decisions are being made, whether shopping for ingredients, selecting meals at restaurants, or cooking at home.   

4. Personalized AI-driven health nudges

When it comes to maintaining a healthy lifestyle, people often struggle due to a lack of motivation. They can also be anxious about whether they can follow through with their new habits, other times they simply find the activities they should do boring and don’t feel rewarded.

A well-timed, customized digital intervention can improve patient experience if it addresses people’s pain points and assists them in taking the correct actions—and doing that without abrupting their everyday lives.

A study examined the barriers to patient adherence to a healthy diet and recommended physical activity and over 27% of participants marked lack of motivation as one of the top three reasons in both cases.

AI-driven behavior design to improve citizens’ health 

Singapore’s Health Promotion Board (HPB) launched the Healthy 365 app that uses the NudgeRank system, an AI-powered behavior design engine. 

It analyses each user’s behavior, demographics, and real-time data from wearables to deliver personalized health nudges, engaging citizens in health challenges and local wellness programs. For instance, if a user who regularly meets their step goals starts to become less active, the system detects this change and sends a motivational nudge, such as a reminder to join a nearby walking challenge. 

This personalized approach led to a 6.17% increase in daily steps and a 7.61% increase in weekly exercise minutes. As users' habits evolve, NudgeRank adapts its nudges, ensuring they remain engaging, relevant, and effective, thereby enhancing the health management experience for over a million Singaporeans.