Image analysis or GCMS analysis in a high throughput manner. Applications of GCMS include drug detection, fire investigation, environmental analysis, explosives investigation, and identification of unknown samples.

Analyse candidate patient data to measure probability of individual resistance to deployed drugs, typically projected over time and across a broader population.

Continual research and observation of the drug's effects on patients after the drug becomes generally available

Predict outcomes from tests. This may mean that fewer experiments are required, thus reducing experimental R&D costs and time to market. With appopriate data and a sensitivity to risk trade-offs it may also mean that a go / no go decision can be made before the cost of a testing regime is initiated.

Predicting prescription adherence with different approaches to reminding patients

Analyse relevant biomarkers to evaluate potential medical applications and outcomes. Identify which genes potentially cause which disorders to simplify diagnosis of patients and provide insights into the functional characteristics of the genetic mutation.

During the initial phase of drug research and development, the target for a new drug treatment must first be chosen. The target molecule will be what the drug compound interacts with to get the intended outcome, often the treatment of a disease. Researchers use deep neural networks to predict molecular level interactions to treat a condition or improve particular functionality

Predict outcomes from fewer or less diverse experiments to reduce R&D costs and time to market - potentially also mitigating cost and loss of life from animal experimentation

Optimise resource allocation in R&D, leveraging multiple data sources (e.g., communications, documentation) to track progress

Predict how successful CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats) will be at editing the targeted genome

Machine learning can be used to speed up the product research and development phase of the chemical industry.

During the initial phase of drug research and development, the target for a new drug treatment must first be chosen. The target molecule will be what the drug compound interacts with to get the intended outcome, often the treatment of a disease.

Predict real-world results from fewer experiments to reduce experimental R&D costs (e.g., new crop testing). This can have a positive environmental impact.

Identifying negative responses (monitor social networks for early problems with drugs)

Analyse relevant characteristics and identify potential individuals to be recruited from a relevant population to serve in drug testing trials. These trials may be different stages in the drug development process.

Leveraging molecule database with metabolic stability data to elucidate new stable structures

content

Identify drug compounds with current regulatory approval which could be used in new ways to treat other conditions. Machine learning assists by searching through existing research literature for known and inferred relationships

Once a drug target has been identified, the drug compound itself must be evaluated for safe use in living organisms before live testing can begin. This includes researching how the drug will be metabolised by the body and identifying potential toxic interactions and side effects.

Identify critical factors to improve R&D efficiency - e.g. to reduce the number of required experiments for research and testing process (an examples might include component testing). Optimising configuration of processes and operations will work better where the parameters under development remain stable.

Analyse the data set (cost / benefit analysis) of potential development projects to assist with prioritisation of research effort and resource allocation in the product development process.

Analysis of clinical outcomes to adapt clinical trial design

Identifying biomarkers for boxed warnings on marketed products. Drug labelling may contain information on genomic biomarkers and can describe issues such as drug exposure and clinical response variability, risk for adverse events, genotype-specific dosing, mechanisms of drug action, polymorphic drug target and disposition genes, and trial design features.

Optimise the testing and manufacturing processes to enable efficient turnaround and throughput for drugs to be trialled during the research and development phase.

Combining medications can produce negative side effects - and potentially mitigate the positive impact. Issues for this include limited overlap case studies, decentralised information and unclear cause and effect. Using AI on appropriate data sets can uncover previously unnoticed correlations. Note that 11% of the US population claim to have used at least 5 medications in a given 30-day period.

Predict outcomes using fewer experiments to reduce experimental R&D costs. Examples would include simpler component testing and using models to minimise the requirement for expensive and time-consuming track testing.

Even with advances in computing power there will come a moment when too much data has been gathered to be economically stored - for example during high spec science experiments. At that stage, in real time, machine learning can be used to help decide which data should be stored for analysis and which deleted.

Data visualisation typically supports better analytics and decision making - collating and standardising data from potentially multiple sources. AI will enable faster and scalable data visualisation with the potential to respond to real time issues. This will be a set of tools increasingly embedded in other applications and use cases.

Monitor and analyse interactions with customers (potentially across all channels ranging from market research to social media to direct contact) to create insights to improve product offering. The aim is to ensure that there is a positive feedback loop in to product development.

Automate generation of articles - for example using national data sets and then producing local variant articles from this. The same can be done for press releases.