With rapid advancements in the artificial intelligence landscape, scientists are using it to make diagnoses and treatments faster and individual-specific. Now, a new study has found a way to use a novel AI model to predict treatment outcomes of tuberculosis (TB) patients, which could lead to personalisation of their treatment.
In the study, researchers from the University of Michigan analysed multimodal data including diverse biomedical data from clinical tests, genomics, medical imaging and drug prescriptions from TB patients.
Through the analysis of patients’ data with different levels of drug resistance, the researchers discovered biomedical features predictive of treatment failure, the university’s press statement revealed. They also found drug regimens effective against particular sets of drug-resistant TB patients. The findings were published in iScience.
"Our multimodal AI model accurately predicted treatment prognosis and outperformed existing models that focus on a narrow set of clinical data," co-author Sriram Chandrasekaran said in the statement. According to the researchers, a key finding was identifying drug regimens that were effective against certain types of drug-resistant TB across countries. This is crucial due to the spread of drug-resistant TB.
The researchers used AI to examine more than 5000 patients. They considered more than 200 biomedical features in the analysis and examined demographic information such as age and gender as well as prior treatment history. Furthermore, if the patients had other comorbidities, such as HIV, then they worked with several imaging features such as their X-ray, CT scans, data from the pathogens, drug-resistance data, as well as genomic features and what mutations the pathogen had.
“It’s really difficult clinically to look at the data all together,” co-author Awanti Sambarey said in the statement. “Typically, you would look at it separately. I think that’s where AI comes in handy. When clinicians look at all of this data, it can be overwhelming. Here, our research is able to identify the most meaningful clinical features,” Sambarey added.
The team also examined the impact of the type of drug resistance present. This means that looking at a specific snapshot of data such as genomic features, one could identify the mutations that the infecting pathogen had and ask what some of the long-term treatment implications are, the statement explained.
The researchers also discovered that certain drug combinations worked better in patients with some types of resistance. Furthermore, they also found that drugs with antagonistic pharmaceutical interactions could result in worse outcomes.
"Using AI to weed out antagonistic drugs early in the drug-discovery process can avoid treatment failure down the line," Chandrasekaran noted in the statement.
The researchers hope that this multimodal approach can help researchers provide more personalised treatments instead of adopting a one-size-fits-all concept.
Previous studies have also used AI models to improve the treatment of various health issues. For instance, a study published online edition of npj Digital Medicine in January found that researchers have developed a unique AI algorithm that can detect sepsis before symptom onset and potentially save lives.
In another study, published in Nature Biotechnology in January, researchers developed an AI tool that can quickly analyse medical images to help clinicians diagnose and better treat cancers that might otherwise go undetected.