Artificial intelligence in drug development is steadily expanding its range of applications.
While in recent years much of the attention has focused on its ability to support the discovery of new molecules, AI can also play a role at other points in a medicine’s life cycle: preclinical research, data analysis, clinical development, manufacturing, or post-marketing monitoring.
This progress opens up new possibilities, but it also raises a fundamental question: how can we ensure that the results produced by an AI system are reliable enough to support scientific and regulatory decisions?
How is AI used in drug development?
Talking about AI in drug development does not mean talking about a single technology.
There are models capable of processing large amounts of information, identifying patterns, making predictions, or supporting specific tasks within scientific and technical processes.
In 2026, the FDA and EMA jointly published guiding principles of good practice for the use of artificial intelligence throughout the drug life cycle. The document reflects precisely that AI applications can extend from the earliest stages of research to clinical development, manufacturing, and post-authorization activities.
Discovery of new drugs
Drug discovery requires analyzing information about diseases, biological mechanisms, potential therapeutic targets, and many candidate compounds.
In this context, certain AI models can help to:
- analyze large biological and chemical datasets;
- identify relationships between different variables;
- prioritize candidates for further research;
- estimate certain properties of a molecule.
This can help guide research, but it does not replace experimental validation. A computational prediction must be tested and analyzed within the scientific context in which it is generated.
Artificial intelligence and preclinical research
Before studying a new drug in people, it is necessary to better understand its activity, its behavior in the body, and its potential risks.
Artificial intelligence in biomedicine can contribute to the analysis of data from experimental studies and to the development of predictive models.
Areas generating particular interest include, for example, toxicity prediction or the prediction of certain biological responses.
However, the usefulness of these tools depends on factors such as the quality of the data used to train them, their validation, and the specific context in which they are intended to be used.
AI in clinical trials
Clinical trials generate large amounts of information: clinical variables, analytical results, imaging data, or information from different sites.
AI in clinical trials can help analyze certain datasets, identify patterns, or support some processes related to clinical development.
But the greater the importance of a decision supported by an AI model, the greater the need to demonstrate that it performs appropriately.
That is why one of the key concepts for regulatory authorities is the context of use.
It is not enough to state that an algorithm works. It is necessary to define what it is used for, with what data, on which population, and with what level of risk.
AI is also reaching manufacturing
Drug development does not end with clinical research.
It is also necessary to establish processes capable of manufacturing a product in a controlled and consistent manner.
Artificial intelligence can have applications related to production data analysis, monitoring certain parameters, or identifying trends within manufacturing processes.
This can be especially relevant in complex production systems, although any tool used must be properly evaluated for the specific purpose it is intended to serve.
The big challenge: reliable models and data
Being able to analyze more data—or do it faster—does not automatically make an AI system a valid tool.
Data can contain biases, fail to adequately represent all populations, or change over time. A model may also behave differently when used outside the environment in which it was developed.
That is why the FDA and EMA highlight aspects such as:
- data quality and governance;
- a clear definition of the context of use;
- a risk-based approach;
- evaluation of model performance;
- documentation and traceability;
- oversight throughout the entire life cycle.
This idea is particularly important in a field such as health: using artificial intelligence does not reduce the need for scientific rigor.
The future of AI in drug development
Artificial intelligence can bring new tools to biomedical and pharmaceutical research, but its value will depend on the ability to integrate it correctly within scientific processes.
Algorithms, data, bioanalysis, clinical knowledge, manufacturing, and regulation will need to advance in a coordinated way.
That is why the future of artificial intelligence in drug development will necessarily be multidisciplinary.
At Qube Technology Park, science, technology, and business share the same Life Sciences ecosystem, creating an environment where different capabilities can come together to drive new health-innovation projects.
