Insurer saves over BRL 16 million in one year with predictive analysis

Facing claims and identifying fraud are constant challenges for insurers striving to protect their financial health. In this market, where each claim impacts finances and amplifies risks, the ability to predict and prevent fraud is indispensable. With the advancement of new technologies like Artificial Intelligence and Machine Learning, new possibilities arise to tackle this challenge.

With the goal of incorporating advanced technologies into its processes, one of the most important insurance companies in Brazil hired act digital. The project involved implementing a predictive model based on AI and Machine Learning, capable of identifying suspicious patterns and improper claims, preventing million-dollar losses.

The solution provided the company with a comprehensive and detailed view of operations, enabling more effective preventive actions. In just the first year after implementation, the insurer saved over BRL 16 million, demonstrating the positive impact and importance of predictive analysis in claims and fraud management.

Moreover, the average effectiveness rate of denials— a metric indicating the proportion of payment rejections for services rendered— increased from around 50% to 70%. These results led the company to project savings of BRL 490 million over five years. Check out all the project details!

Client: over a century in the insurance market

Founded in 1895, the company is one of the largest insurers in Brazil. In 2007, it became a publicly traded company, strengthening its relationship with investors and the market. Currently, the group serves over 7 million clients, offering insurance products in various sectors, including health, dental, life, and pensions, as well as investments.

Through these products, the company fulfills its mission of ensuring a better life for people by providing support, security, and autonomy at all times. Its solutions focus on ensuring physical, emotional, and financial well-being.

With more than 4,000 employees, the insurer has a comprehensive network of specialists, including brokers, doctors, consultants, and other professionals in the insurance and investment sectors.

Challenge: increasing accuracy in claims, denials, and fraud analysis

The company faced financial losses due to fraud, claims, and denials paid improperly. For example, denials, which are rejections of payments for services rendered, when not identified, generated avoidable costs. The issue was that the company’s analysis system did not provide the necessary precision to reduce these losses and optimize processes.

The insurer already had a system in place, but it was outdated and did not meet operational needs. For instance, the deviations detected by the tool were limited, and some crucial analyses were not covered (such as those related to materials and medications that could be denied).

Expected objectives

• Reduce financial losses.
• Predict risks of fraud, deviations, and abuse.
• Increase the variables and situations analyzed by the system.
• Enhance the company’s analytical and predictive capabilities.

Solution: Enhanced system using AI and Machine Learning

To address the challenges, a predictive model was developed to identify and prevent denials, structuring a robust database and integrating data from multiple transactional sources. The solution included data visualization and continuous monitoring of the predictive model’s performance, enabling more efficient preventive actions.

Methodologies

Methodologies such as Design Thinking, Human-Centered Design, and DesignOps were adopted to underpin the design process. Additionally, the Double Diamond tool was used to conduct UX research and develop prototypes.

Technologies used

• Google Cloud Platform (GCP): utilized for data structuring and governance.
• Python: implemented to develop the machine learning predictive model.
• Tableau: used for data visualization.
• Miro: tool for conducting and documenting all UX research.
• Adobe XD: utilized for prototyping the solution.

System changes implemented

• Deviations: the tool, which previously only evaluated two types of deviations, now considers eight variables.
• Materials and medications: the system began analyzing situations related to the use of materials and medications (which could also lead to denials).
• Expanded predictive analysis: the software extended predictive analysis to all types of accounts, covering both administrative and technical areas, whereas it was previously limited to the latter.
• Mass denial: a functionality that was not previously offered by the system.
• Dashboards: data panels with information that can be monitored by the teams.

Results: Million-dollar savings from claims and deviations

The implementation of the solution brought significant improvements to the company, optimizing processes and reducing losses. The technical and administrative analysis of denials was automated, allowing for faster fraud detection. As a result, operational efficiency increased, leading to a significant reduction in fraud and denials, generating savings for the insurer.

Main results

• Process automation: improved technical and administrative analysis of denials.
• Quick fraud recognition: increased agility in fraud detection.
• Operational optimization: reduced manual steps in the analysis of denials.
• Financial savings: significant reduction in fraud and denials, generating savings for the company. In the first year, the company saved BRL 16.6 million. The five-year projection estimates that the savings will reach BRL 490 million.
• Greater effectiveness: the average effectiveness rate in denial analysis increased from around 50% to 70%.

How we can help your company

act digital’s services ensure security and precision, offering innovative solutions through advanced technologies. Effective data management and efficient operational processes help insurers adapt to the new digital market.

We deliver innovation through a modern and digital business ecosystem, monitoring and analyzing data in real time. We create predictive models that anticipate fraud and undue claims, and offer personalized solutions tailored to the specific needs of each client. Check out some of our main approaches in this sector!

• IoT and Machine Learning: We integrate connected devices and advanced algorithms to collect and analyze data in real time, identifying patterns and preventing risks.
• Cloud Computing: We utilize cloud platforms to securely and efficiently store and process large volumes of data.
• Artificial Intelligence: We develop AI models to predict and mitigate fraud, improving decision-making and risk management.
• Strategic data management: We implement advanced data management practices, ensuring the integrity, security, and utility of information for strategic decision-making.

Does your company need an advanced solution in predictive analysis? We can help. Contact our specialists!