How to maximize artificial intelligence in the healthcare field

2023-09-25 08:53:26

In recent years there have been great advances in the field of Artificial Intelligence (AI) that allow us to verify its potential in personalized precision medicine. However, there are still “many challenges” to achieve the implementation of these systems.

Image from the Roche Institute Foundation report.

It is one of the main conclusions of a new Anticipating Report on Applications of Artificial Intelligence in Personalized Precision Medicinepublished by the Roche Institute Foundation that addresses the path that these tools have ahead for their proper integration in the healthcare field

The document outlines the potential that Artificial Intelligence has in personalized precision medicine due to the improvements it allows in research and clinical practice, not only for professionals and patients, but also for the health system.

A paradigm shift in clinical practice

In this way, the report emphasizes that AI is a “promising tool” with a view to automating problem solving, improving decision-making capacity and more accurately characterizing health and disease states.

Its ability to analyze large volumes of information as well as find “complex patterns” will predictably lead, according to the document, to significant improvements, from the prevention and prediction of disease risks, early detection and diagnosis, to monitoring and treatment. of diseases.

Report graph.

“Not only can it represent a paradigm shift in routine clinical practice, but it will also be possible to apply it in biomedical research with drug discovery, or other applications that are not yet well developed, such as its potential for education and training in the field.” biosanitary”, he emphasizes.

And he maintains that the establishment of “synergies” between Artificial Intelligence and Personalized Precision Medicine It’s fundamental.

He insists that there is “great potential” in the diagnosis, treatment, detection and classification of diseases, but also for the improvement of processes and the safety and quality of life of the patient, “always under the supervision of health professionals.” , he points out.

Your contribution to healthcare practice

In healthcare practice, it can contribute, according to the report, to adopting actions that contribute to reducing the burden of disease in health systems and “optimal use” of health resources.

In fact, there are tools, for example, to help understand, in the context of a pandemic, the impact of the massive implementation of telecare, the closure of Primary Care consultations or the provision of more resources to emergency services, on the care of different pathologies.

And AI can also be used, among other things, to develop disease risk prediction models that, until now, were based on traditional mathematical models.

AI challenges around precision medicine

Regarding precision medicine and AI, there are technical challenges related to the large amount of data that is currently generated and managed, according to the Roche Institute Foundation report, which are as follows:

The limited availability and access to quality data: Its generation and storage is usually not standardized or automated and, sometimes, access to the data generated by different agents is not public, which makes it difficult for systems to use said information appropriately.

The available data may “not be representative and lead to biases”: Databases do not usually have all the possible existing data, so the information they collect “is incomplete.” An Artificial Intelligence model trained with data from these databases could amplify bias and achieve erroneous results. This means that it can recommend “unfavorable decisions” towards a particular group of people characterized by age, gender, race, geographic or economic level.

The reliability of the solutions and results obtained by AI: New systems such as ChatGPT and Large Language Models make visible, the document points out, the path that remains to be followed in their development in order to guarantee their reliability. They can make “numerous” errors, which could have “fatal consequences in the field of medicine.”

The need to invest time and resources: Computer equipment and systems for data management, as well as tests to validate the systems, may require large investments of money and time. However, the application of artificial intelligence is expected to contribute to the advancement of personalized precision medicine. Also that reduce your costs, improve the quality of care and contribute to the sustainability of the health system.

More implementation challenges in clinical practice

And regarding the challenges of implementation and translation to clinical practice, the report highlights the explainability and interoperability of the information generated by these systems.

In this sense it abounds that “It is very complex” for AI systems themselves to explain how they arrive at the results they obtain. This represents a limit to the understanding and understanding of the systems by healthcare workers.

It also highlights the difficulty in carrying out evaluations and validations of the systemsin addition to the regulatory environment, which is still developing, despite the fact that the existence of these AI systems in health dates back to the 1970s.

And among the challenges is also the problem of the vulnerability of systems that, like any software, can be hacked.

Recommendations to promote your potential without forgetting your limitations

Thus, the report proposes a series of recommendations to maximize Artificial Intelligence in the healthcare field, addressing its possible limitations:

Standardize the collection and management of health data according to standards to ensure “the quality, homogeneity and interoperability” of the same and avoid biases.

Appropriately evaluate and validate AI systems in the face of “lack of transparency” due to the type of data that feeds the models, the calculations made by the model or even the overfitting of these models to the conditions of the environments in which they are developed

Promote the incorporation of AI in clinical protocolsclinical practice guidelines and consensus for use in healthcare practice.

Promote the training of health professionals in this field and establish the specialty of biomedical informatics.

Spread and raise awareness about the potential and limits of Artificial Intelligence in medicine.

Design policies privacy, confidentiality and data protection.

Establish a specific regulation for the use of AI in the health field.

Create a competitive national industry of Artificial Intelligence.

.