Mélanie Guirette
PhD Candidate, Nutrition Epidemiology and Data Science, Friedman School of Nutrition Science and Policy, Tufts University
Final Presentation
Final Project
How to use AI to fight IBS
Project Description
The circulating metabolites in our system, known as the metabolome, is highly influenced both by the environment (including diet, medicine, light and pollutant exposure), and the individual phenotype (our genes, our metabolism) However, the key determinants of most metabolites are still poorly understood.
Irritable bowel syndrome (IBS) is one of the most common gastrointestinal disorders worldwide, characterized by chronic abdominal pain or discomfort and altered bowel habits.
To date, the exact pathogenesis of IBS remains elusive, but is clearly multifactorial, including environmental and host factors
We propose the development of a machine-learning algorithm that predicts metabolite levels across different observational cohort studies using their data availability on genetics, gut microbiome, metabotype data, anthropometrics, and lifestyle factors.
We propose the development of a machine-learning algorithm that predicts metabolite levels across different observational cohort studies using their data availability on genetics, gut microbiome, metabotype data, anthropometrics, and lifestyle factors.
The two main ethical concerns in this project include 1)Consent and data privacy, since we are using previously collected data for research that may not have been intended during data collection. 2)And the obligation to protect persons from harm by maximizing benefits while minimizing potential harm(s).
To address these concerns, we plan using a human centric design so that the discoveries in health of our OMICs research will benefit the participants in a way that outweighs their risk of data breach which will be minimized using ethical, IRB approved practices in data sharing, management, and security that are HIPAA compliant
Our project will link the ontologies of metabolomics, the microbiome, and IBS, where:
Metabolomics is the large-scale study of metabolites that are found circulating in our system at varying levels depending on the biofluid (blood, urine, or fecal), and uses either untargeted or targeted techniques depending on the purpose of the study – untargeted to qualify all the metabolites present; targeted to quantify the amount of known metabolites present
Metabolomics allows for the discovery of gut derived metabolites produced by the microbiome, which through association studies can lead to biomarker discovery of conditions such as IBS
Which as we previously mentioned, is an umbrella condition which is broadly organized by different subtypes depending on symptoms: d for diarrhea, c for constipation, m for mixed, or u for unknown.
Taking a closer look at the causal diagram that links the fields of our research. Lets break it down.
We know dietary intake affects IBS symptoms, but we don’t know very well how.
Our hypothesis is that IBS symptoms are primarily linked to our diet through the gut microbiome, and the activity of the gut microbiome can be measured by the metabolites it produces and releases into our system. Our AI model will use these metabolite levels so that it can start predicting what dietary triggers are causing IBS symtoms
However, there are external factors outside of this relationship that can confound the links we are trying to assess, and should be taken into account. One of these factors illustrated here is genetics, but it can also be other lifestyle factors like physical activity, medication use, air pollutions, etc; all of which will me taken into account in our model depending on what data is available
This is a complete concept map of the use of metabolomics in IBS, but we can focus on a subset of sections to link this research to policy:
First, Use of metabolomics to elucidate symptoms of IBS can inform healthcare policy and public health in the realm or personalized medicine and preventative measures
Second, this research informs clinical practice by using AI-driven predictions as a diagnostic tools to identify IBS subtypes and appropriate dietary recommendations as treatment
Finally, overall applications of this technology promotes its R&D at the national level, where efforts to standardize data collection and support big data initiatives become possible
Hi Melanie,
This is a fascinating project with the potential to make significant strides in IBS research. Your approach could provide crucial insights into dietary triggers for IBS, which would be a valuable addition to the field. Excited to see the outcomes of this study!
Hi Melanie,
This is an interesting use case that addresses an important gap in the literature! It seems that leveraging the ability of AI models to detect patterns in large volumes of data is well-suited to answer your research questions. It will be interesting to see how these tools can filter through and synthesize data on genetics, gut microbiome, metabotype data, anthropometrics, and lifestyle factors compared with how an expert in the field would do so.
Given the umbrella usage of IBS as a term, I think that being able differentiate the presence or, lack there of, of certain metabolites will be enlightening and even more differentiating for the field of IBS research, diagnosis, study design, and treatment. I am excited to see where this research could possibly lead.
Hi Melanie, Thank you for such a complete and straight forward presentation. I think your research could also use the idea of digital twins as a next step to see how various interventions (mostly dietary) woud have an impact on quality of life for those suffering from IBS. I am also not overly familiar with the subtypes so I am wondering if someone stays with this same type? Or can they have C-IBS and then certain environmental or dietary triggers could switch to D-IBS? Overall, this is very clear and I hope to read about your methodology some day soon!
Hi Melanie, I really enjoy reading your work! It’s particularly impressive how you plan to tackle the ethical concerns associated with using previously collected data. Your commitment to maintaining a human-centric design and ensuring that the benefits to participants outweigh the risks sets a high standard for research integrity and participant respect.
Hi Melanie, I think this is an incredibly useful and fascinating use case! We know so little about the gut microbiome and work like this would be a game changer! I am really excited to see where you take this. Awesome work!
Hi Melanie, your project’s integrating metabolomics and AI to understand IBS is impressive! Can’t wait to see this approach.
Hi Melanie! You managed to make a very complex topic more straightforward to understand. I agree with others that the complexity and detail of this topic are well-suited for AI. IBS is such a general topic that creating specificity will be invaluable, but it will also require a human-centric approach. I appreciate your attention to ethical considerations here. Well done, and good luck!