AI in Public Health: Advances, Gaps and Opportunities of the Brazilian Artificial Intelligence Plan

Alma Mater Cosméticos
Jun 15
15 min read

Artificial intelligence (AI) is emerging as a transformative ally in public health, capable of improving diagnoses, optimizing resources and personalizing treatments. Recognizing this potential, Brazil recently launched the final version of the Brazilian Artificial Intelligence Plan (PBIA) 2024–2028 , also called “ AI for the Good of All ”.

With a planned investment of R$23 billion by 2028, the PBIA seeks to position the country among global leaders in AI, guiding technology to solve social challenges and improve the well-being of the population. One of the central focuses of the plan is the health sector – especially public health and the Unified Health System (SUS) – given the vast amount of patient data and the unique genetic diversity of the Brazilian population.

In this article, we assess the main points of the PBIA aimed at innovation in health, identifying gaps in the plan and proposing strategic paths. It also presents international examples (United Kingdom, Estonia, Singapore, among others) that can inspire policies and initiatives for Brazil to leverage its potential in AI in health.

Health Innovations in the Brazilian Artificial Intelligence Plan

PBIA emphasizes the modernization of the SUS through AI , with seven immediate impact actions dedicated to public health. These initiatives aim for results in 12 months and leverage the SUS’s extensive database.

These include: Spoken Medical Records in the SUS (use of AI to automatically transcribe medical teleconsultations), AI for Medication Purchasing Decisions (assistance in the management of public drug purchases), Diagnostic Optimization in the SUS (AI for faster and more accurate diagnoses in critical conditions such as stroke, pneumonia, cancer and tuberculosis), AI in Oral Health (technology to improve dental services and detect oral cancer), Detection of Anomalies in Procedures (identification of irregularities in billing and hospital procedures), Support for the Management of Judicialization Processes (use of AI to manage healthcare lawsuits, preventing litigation) and Elderly Well-Cared For (AI platform to promote care for the elderly population).

These fronts, in addition to signaling PBIA's commitment to applying AI directly to SUS services, aim to achieve gains in administrative efficiency and clinical improvements. The expected impacts of these innovations include faster and more assertive diagnoses (for example, speeding up stroke detection), intelligent decision support in purchasing and logistics of medicines, identification of fraud or errors in procedures, as well as improvements in specific areas such as oral health and elderly care.

In short, the plan seeks to take advantage of the scale and reach of the SUS so that AI solutions can benefit millions of Brazilians. The Minister of Science, Technology and Innovation, Luciana Santos , highlighted that health is one of the sectors most covered by the PBIA and that the Covid-19 pandemic has highlighted the urgency of reducing external technological dependencies.

In fact, R$435 million has been allocated for these immediate actions in nine priority sectors – health being one of the main ones – with short-term deliveries. In addition, solutions in public health and SUS are listed by PBIA as one of the nine areas in which Brazil can exercise strategic leadership in AI, reinforcing the priority of health in the national innovation effort.

SUS Data and Genetic Profile: Brazil's Differentiators

A notable asset of Brazil for AI in healthcare is the SUS itself, which for decades has collected data from a population of over 200 million inhabitants, covering enormous ethnic and regional diversity. According to a study by NIC.br/Cetic.br , the SUS represents a competitive advantage in obtaining health data for AI, given the capillarity, scale and diversity of the Brazilian population served.

The national databases of the SUS – if well integrated and managed – provide rich raw material for algorithms to reliably train, reflecting a wide variety of real-world clinical cases. This perception is corroborated by experts, who see enormous value in the data collected by the SUS and consider its use an important step towards integration and intelligence in health systems. Recent initiatives by the Ministry of Health, such as the National Health Data Network (RNDS) , already aim to standardize and connect electronic medical records nationwide, paving the way for large-scale AI applications by enabling the secure exchange of information between points of care.

Another strategic asset is the diverse genetic profile of the Brazilian population . The result of centuries of interbreeding, the Brazilian genome has unique variants that can support scientific discoveries and precision medicine applications. A recent study sequenced the DNA of 2,723 Brazilians from various regions, identifying 8.7 million previously unmapped genetic variants . This demonstrates how unique and little explored the national genomic heritage is – a “genetic blackout” that Brazilian researchers are beginning to illuminate.

In 2020, the Genomas Brasil Program was launched with the ambition of mapping 100,000 Brazilian genomes, placing the country on the population genetics map. Although the PBIA does not address the integration of AI and genomics in depth, this convergence is promising: algorithms can cross-reference genetic data with clinical histories to predict disease risks and guide personalized preventive interventions.

In advanced countries, such initiatives are already yielding concrete results. In Estonia , for example, 20% of the population has had their genome sequenced and linked to national health records (including cancer databases, causes of death, hospital data, lifestyle habits, etc.). With this integrated database, the Estonian government has been able to reduce the age at which mammography screening begins for women genetically predisposed to breast cancer and to diagnose rare diseases that were previously impossible to identify at an early stage.

Brazil can be inspired by this model, joining efforts from PBIA with genomic programs to leverage AI-driven precision medicine – always respecting ethics, informed consent and the protection of sensitive data.

Gaps and Challenges of PBIA in the Health Area

Despite its merits, the PBIA has important gaps in health care . Experts warn that without addressing these blind spots, Brazil could repeat past mistakes in innovation policies. Below, we highlight some omissions and challenges that need to be addressed:

Data Infrastructure and Interoperability: The plan recognizes the need for sovereign cloud and supercomputers, but does not detail how to prepare the specific infrastructure for the health sector. Today, much of the SUS data is still fragmented across multiple systems (hospitals, basic units, laboratories) with varying quality standards. Integration via RNDS is underway, but deficiencies in data quality and integration persist and may compromise the proper functioning of AI.

It will be necessary to invest in the standardization of electronic medical records, connectivity of health units (especially in remote regions) and in secure repositories for large volumes of clinical data.

Furthermore, healthcare devices and equipment need to be able to collect and transmit accurate data (e.g., good-quality radiology images for computer vision algorithms). Without this robust technological base at SUS points of care, the proposed AI solutions will hardly achieve national impact.
Specific Regulation and Safety: The PBIA proposes a national regulatory framework for AI by 2027 with general ethical guidelines, but does not address specific regulations for AI in healthcare. This is a critical point: how to certify and approve medical algorithms for clinical use? Who is responsible in case of an error in AI in diagnosis or treatment?

Currently, agencies such as ANVISA still lack dedicated frameworks for AI software in healthcare. Cetic.br ’s research highlighted regulatory issues specific to the healthcare sector as one of the main barriers to the effective adoption of AI. It will be necessary to develop technical guidelines and regulatory sandboxes to safely test medical AI solutions, as well as update telehealth and digital health standards to incorporate AI systems.

Another aspect is to align the PBIA with the LGPD (General Data Protection Law) , which classifies health data as sensitive and imposes requirements for its use. The plan mentions principles of data sovereignty, privacy and ethics , but lacks operational details on how to ensure confidentiality of patient data on a large scale. Leaks of health information or misuse of AI can cause serious harm – a risk highlighted by experts.

Therefore, reinforcing information security (encrypted storage, access control, audits) and anonymization/seudonymization of data for research are mandatory measures when implementing PBIA in healthcare.
Collaboration with Startups and Innovation Ecosystem: The PBIA plans to allocate almost 60% of the resources (around R$13.8 billion) to business innovation , including support for AI startups. However, the plan does not explain how these startups – especially national healthtechs – will be integrated into the needs of the SUS.

There is a gap in coordination between innovative solutions emerging in the market and large-scale adoption by the public system. Without agile mechanisms for contracting, accelerating, and testing new technologies in real SUS environments, we run the risk of innovations being restricted to pilot projects. It would be important for the government to create specific challenges and calls for proposals for AI in healthcare , encouraging startups to solve SUS problems (for example, detecting epidemics, optimizing waiting lists, or supporting clinical decisions in primary care).

Furthermore, innovation hubs such as the Digital Hospital in Pernambuco or initiatives in university hospitals could serve as testing grounds to validate algorithms before a nationwide expansion. The lack of mention of these arrangements in the PBIA suggests a missed opportunity to better connect the plan to the strategies of the Ministry of Health and to the public-private partnerships already underway in the health-industrial complex.
Training Human Resources in Healthcare: The PBIA includes actions to train 20,000 professionals/year in AI by 2028, but does not focus on training healthcare teams to deal with AI. To reap the benefits, doctors, nurses, managers and technicians of the SUS need to be trained in the safe and effective use of these tools. It is necessary to promote digital literacy and trust in AI among professionals, showing that technology is here to support and not replace.

Without this preparation, even excellent tools may encounter resistance or inappropriate use. Continuing education in digital health should go hand in hand with the implementation of technological solutions – a point that the plan does not detail.

Furthermore, there is a lack of policies to retain AI talent in Brazil (the plan aims to retain talent, but faces global competition), as well as incentives for researchers to work specifically on national health problems. Integrating universities, medical research centers and companies in joint projects would be a way to mitigate this gap in applied training.

In summary, the PBIA offers a promising direction, but it needs to be complemented by concrete implementation measures in the health sector. Infrastructure, regulation, data protection, integration with startups and training are pillars that require detailed action plans. Their absence in the document indicates that the success of the strategy will depend on complementary policies from the Ministry of Health and other actors, otherwise good intentions will not translate into real changes in citizen care.

Strategic Paths to Leverage AI in Healthcare in Brazil

Given the points raised, there are a series of strategic actions that different actors can adopt so that Brazil realizes its AI potential in health:

Governance and Integrated Planning: It is essential to align the PBIA with a national digital health strategy . It is recommended to create an interministerial committee linking MCTI and the Ministry of Health to coordinate investments and specific goals (e.g., reducing waiting times or improving clinical indicators through AI). This committee could define priority use cases in the SUS and monitor their implementation, ensuring that AI solutions meet the needs of the public system. In addition, the participation of ethics councils and civil society in guiding these policies should be institutionalized, providing transparency and social acceptance to innovations (especially when they involve sensitive citizen data).
Strengthening the Health Data Infrastructure: Accelerating the expansion and improvement of the RNDS is an urgent step. All levels of care (from health centers to tertiary hospitals) need to be connected and feeding a unified repository of clinical data, updated in real time. Secure cloud computing technologies – aligned with the sovereign cloud proposal – must host this data, ensuring performance for large-scale algorithm training.

Brazil should consider creating data lakes for research, with anonymized information from millions of patients, available to universities, startups and companies under responsible use agreements. Similar initiatives exist in other countries: in the United Kingdom , for example, the NHS has developed secure environments for researchers to access patient data in anonymized form, boosting AI studies without violating privacy. Here, partnerships with institutions such as Fiocruz, Datasus and teaching hospitals can enable similar environments, in which predictive models are trained with genuine Brazilian data – which increases effectiveness and reduces imported biases.
Clear Regulatory and Safety Frameworks: In anticipation of the general AI law, Brazil can create specific guidelines for AI in healthcare . This includes establishing criteria for approval of software as medical devices that use AI (defining requirements for accuracy, clinical validation, post-implementation monitoring), as well as protocols for accountability and explainability of algorithms used in medical decisions.
One inspiring example is the UK , where regulators and the NHS have developed codes of practice for AI in healthcare and a regulatory guidance service for developers. Singapore , too, has placed an emphasis on explainable and safe AI as part of its national strategy, recognising that doctors and patients will only trust the technology if they understand its fundamentals and limitations.

Brazil must also strengthen data protection : increase training for SUS managers in LGPD, invest in hospital cybersecurity (health data is an increasingly targeted target) and implement leak detection tools. Additionally, privacy by design practices – such as advanced anonymization and granular consent – need to be incorporated into the development of solutions from the beginning.
Public-Private Partnerships and Startups: To harness the country’s innovative talent, it is strategic to create bridges between the NHS and the startup ecosystem . The government could launch programs along the lines of the UK’s NHS AI Lab , which funds and tests AI innovations in NHS hospitals. In fact, the NHS has an AI lab that has already invested more than £140 million in projects and created an AI Award to accelerate promising healthcare solutions.

In Brazil, an equivalent could offer thematic calls for proposals (e.g., AI for Primary Care, AI for Hospital Management) where startups and research groups compete, receiving resources and access to non-sensitive data from the SUS for development. In addition to funding, these partnerships should include controlled testing environments : for example, implementing a triage algorithm in reference emergency rooms and measuring results before scaling.

Large companies and investors also have a role to play – technology corporations can provide cloud infrastructure or AI expertise, while impact investors can support startups aligned with the needs of the SUS. The goal is to foster an AI health economic-industrial complex , where local innovation solves local problems and then reaches the world. This synergy between government, startups, universities and the private sector creates a virtuous cycle of open innovation.
AI Training and Culture in Healthcare: No technology can thrive without skilled people. Healthcare workforces need to be trained at all levels. This ranges from incorporating data science and AI content into medical, nursing, and public health curricula to hands-on training for practicing professionals.

Radiologists, for example, must learn to interpret the results of image analysis algorithms; managers need to know how to use predictive tools to allocate resources; and IT professionals in the SUS must master new AI platforms.

At the same time, technology experts should also be trained in the specificities of the healthcare sector – promoting multidisciplinary teams that speak both the medical and algorithmic languages. A culture of continuous learning should be encouraged, with the sharing of success stories and lessons learned from AI projects in hospitals and healthcare units.

Furthermore, engaging patients and the general public is crucial: communication campaigns can shed light on how AI improves care (e.g. by reducing waiting times or preventing errors), building trust. Digital citizenship initiatives – educating the public about their data rights and how AI tools make decisions – also empower users and increase transparency.

International Inspirations for AI in Healthcare

Several nations are already reaping the benefits of AI applied to healthcare, offering valuable lessons for Brazil. Below, we highlight some relevant examples:

United Kingdom: With one of the most digitally advanced public health systems, the UK launched the NHS AI Lab in 2019, dedicating around £250 million to integrate AI into the healthcare system.

The results are tangible: today, 90% of NHS stroke networks use AI , which has halved the average time to treatment and tripled the proportion of patients who regain function after stroke. Algorithms such as Brainomix e-Stroke analyse brain scans in seconds, telling doctors the best course of action and speeding up critical decisions.

On another front, the NHS has funded the deployment of AI imaging for early detection of lung cancer across 64 hospitals, with £21 million dedicated to tools that read X-rays and CT scans more quickly.

The UK government has also developed a Code of Practice for AI in Healthcare and involves healthcare professionals in the design of solutions. The key takeaway is that targeted investment and clear governance can safely scale AI, generating tangible benefits for patients (reduced waiting times, more accurate diagnoses) and for the system (optimized resources).

Brazil, with a nationwide SUS, can be inspired by this AI laboratory model linked to the public system, adapting it to our dimensions and epidemiological context.
Estonia: This small Baltic country has become a global leader in health data and genomics . Since 2000, Estonia has invested in digital infrastructure (99% of prescriptions are electronic) and created a National Biobank that has collected DNA from around 200,000 citizens – an impressive 20% of its population.

Estonia’s unique move has been to fully integrate genetic data into the healthcare system : the national electronic medical record incorporates DNA information, allowing doctors to access, for example, a patient’s genetic report during a consultation. With the help of AI, this data is cross-referenced with clinical and lifestyle histories to generate individualized risk profiles .

Thus, the country was able, for example, to identify women with mutations associated with breast cancer and to anticipate preventive protocols , such as earlier initiation of mammograms. Rare genetic predispositions were also discovered, enabling early interventions in asymptomatic patients.

It is important to note that all of this was done with ethical rigor: Estonians implemented broad consent, governance with independent ethics committees, and a citizen portal (“My Genome”) where each participant can view and manage their personal data. Estonia demonstrates that, even with limited resources, it is possible to innovate by integrating AI, health big data , and genomics to improve public health.

For Brazil, which has a much larger and more diverse population, the Estonian example reinforces the viability of national genomics and personalized medicine programs. At scale, we could target at-risk subgroups with preventive measures (as Estonia has done) and boost research into diseases prevalent in our multiethnic population – a field where AI is essential for analyzing massive amounts of genetic and clinical data together.
Singapore: Known for its Smart Nation strategy, Singapore has made healthcare a priority in its National AI Strategy . The Asian country's focus is on preventing and managing chronic diseases (such as diabetes and hypertension) using AI on a large scale.

One of the flagship projects is the development of a personalized risk score for each citizen, combining clinical data, imaging tests, lifestyle habits and even genomic information. AI analyzes this data to assign a calculated risk of the individual developing cardiovascular or metabolic complications, for example, enabling proactive interventions.

At the same time, Singapore is implementing AI into self-care tools: apps that remind patients to take medication, monitor diet and exercise, and advise when to seek medical help.

One success story is the SELENA+ system, adopted nationwide for automated analysis of retinal images. Initially designed to detect diabetic retinopathy with 70% greater efficiency than human evaluators, SELENA+ is now being expanded to predict cardiovascular risks from signals in the eyes.

Singapore’s experience shows the importance of having clear goals (in this case, reducing the burden of chronic diseases) and integrating AI at both the clinical level and in patient engagement in their care. For the SUS, which deals with a high prevalence of hypertension, diabetes and other chronic conditions, this approach could inspire national AI-driven preventive health programs – for example, using predictive models to monitor hypertensive and diabetic patients registered in Primary Care, avoiding unnecessary complications and hospitalizations.

Other countries are also offering learning opportunities, from Canada , which is investing in centers of excellence in AI applied to mental health, to Israel , which has made anonymized databases from its HMOs available to startups to develop innovative medical solutions. What successful initiatives have in common is a combination of long-term strategic vision , sustained investment, adequate regulatory frameworks, and a collaborative ecosystem between government, academia, and companies. This synergy is exactly what Brazil needs to cultivate around PBIA and the digital transformation of the SUS.

Conclusion

Brazil faces a unique opportunity to unite its vocation in public health – consecrated by the SUS – with the artificial intelligence revolution. The Brazilian AI Plan lays the foundations and demonstrates political will to make AI a tool for the common good.

In the healthcare sector, the potential benefits range from life-saving diagnostic support systems that are fast-tracked to more effective preventive health policies driven by massive data and advanced algorithms. However, reaping these rewards requires more than just paper: it requires strategic execution . This includes investing in the right infrastructure, adjusting legal frameworks, rigorously protecting citizen privacy, encouraging startup innovation, and empowering people to work side-by-side with intelligent machines.

Fortunately, we are not starting from scratch. We have the SUS as a pillar , an emerging scientific and startup community in digital health, and international examples that pave the way. If the government, universities, companies and civil society act together – each contributing expertise and resources – Brazil can position itself at the forefront of AI in health, developing solutions at home for our challenges and exporting this knowledge to the world. The aim is to enable a new era of more preventive, personalized and efficient health, without abandoning the principles of universality and equity that govern the SUS.

Ultimately, success will be measured by the impact on people’s lives: patients seen in time thanks to an AI alert; communities where smart epidemiological surveillance prevents outbreaks; healthcare workers freed from repetitive tasks to focus on human care; seniors safely monitored at home; medicines reaching where they are needed most, when they are needed.

These scenarios are within our reach if we persist on the path of responsible innovation. The PBIA 2024–2028 is a call to action – it is up to us as a nation to turn its potential into reality.

At the Brazilian Institute of Innovation in Health - IBIS , we believe that artificial intelligence can be a strategic ally in transforming healthcare in Brazil, as long as it is connected to science, ethics and the real needs of the public system and the population. We work alongside startups, companies, researchers and policy makers to build bridges between technology and impact. If you are developing innovative healthcare solutions or want to support this transformation, we are open to dialogues, partnerships and joint projects.

For those who wish to learn more, the Brazilian Artificial Intelligence Plan (final version, 108 pages) is available for download on the MCTI/CGEE website. Access the official document at this link.