Dr. Kirsten Bischof & Dr Zain Khalpey
Go with your gut: Could Artificial Intelligence (AI) lead research in the field of the Human Microbiome (HMB) and be the tool that turns back the clock on Alzheimer’s (AD) and other neurodegenerative disorders?
Introduction
“Let food be thy medicine, thy medicine shall be thy food” – Hippocrates
The saying “You are what you eat” is a timeless reminder of our dependence on a healthy diet. However, with technological advances in AI, scientists have made remarkable strides in comprehending the human microbiome. The microbiome plays a critical role in understanding and treating a range of pathologies, from Ulcerative Colitis and Cardiovascular disease to Alzheimer’s and Autism Spectrum Disorders. It is often called the “forgotten organ” because it is vital to essential physiological processes, from metabolism to immunity. Despite the challenges posed by genetics, gender, diet, geography, and other heterogeneous features, scientists have found breakthroughs in identifying the best recipe for a healthy human microbiome. The microorganisms that comprise the microbiota generate metabolites that are believed to be the key ingredient in the phenotype of several disease states. This is particularly exciting because it opens the door to discovering new biomarkers, reducing the invasiveness of diagnostic testing, and potentially preventing the evolution of numerous crippling illnesses. We can envision a future where a simple stool sample could predict the onset of diseases like AD and PD, and something as innocuous as a probiotic could prevent or halt the progression of that disease. In light of this information, it is clear that “going with your gut” is an excellent idea, and we have realized that our “gut feeling” has more to do with our overall health than we ever imagined.
What is the Human Microbiome, and what role does it play in neurological disease?
Without a doubt, the human body serves as a host to an incredibly diverse range of microorganisms that live in perfect symbiosis with it. The gut microbiome, primarily composed of bacteria but also includes protozoa, viruses, and archaea, plays a crucial role in regulating essential physiological functions like the maturation of the innate and acquired immune system. It’s intriguing to note that there are more bacterial cells present in the human body than human cells, and the genome of the human microbiome is 100 times greater than the human genome.

The gut microbiome:
- Performs a wide range of crucial functions, such as promoting efficient digestion and absorption of nutrients, which is vital for creating essential metabolites, including short-chain fatty acids.
- Additionally, it plays a crucial role in supporting the maturation of the digestive tract
- and acts as a formidable barrier against pathogens and toxins.
- Moreover, it is instrumental in developing the immune system.
- It is essential in synthesizing vital vitamins such as B6, B3, B9, and B12.
The gut and brain are connected through a complex network of signaling pathways called the gut microbiome-brain axis. This network involves the communication between the microbiota, microorganisms in the gut, and the brain. Dysbiosis, which refers to an imbalance in the functioning of the microbiome, can cause dysregulation in these communication pathways. This can lead to increased permeability of the blood-brain barrier, which in turn can cause neuro-inflammation. These conditions can contribute to pathological diseases such as Alzheimer’s disease, Parkinson’s disease (PD), depression, and Autism Spectrum Disorders (ASD).
The gut microbiome mainly comprises Bacteroides and Firmicutes, which comprise 90% of its phyla. However, the traditional microbial techniques are limited to culturing less than 70% of the microbiome microorganisms. Thanks to the improvements in metagenomics and next-generation sequencing, scientists can now study and understand the microorganisms that create the microbiome. This new knowledge helps them to investigate the potential role that these microorganisms could play in neurodegenerative diseases.
Why are we interested in it?
- Drug metabolism relies heavily on it, making it a prime target for the biotech industry.
- It has the potential to unlock preclinical diagnoses and treatments for neurodegenerative diseases such as Alzheimer’s and Parkinson’s.1
- Currently, there are few effective therapies for modifying the progression of Alzheimer’s.
- With 30-400 trillion microorganisms generating terabytes of data, it’s an ideal subject for AI because it thrives on big data.
- As we age, our dignity becomes increasingly important to us. It’s not about losing physical abilities but about feeling disconnected from our memories, thoughts, and relationships. Losing touch with these essential parts of our humanity can profoundly impact our well-being and quality of life.
- Research suggests a “healthy” HMB is essential for overall well-being and longevity, as evidenced by the centenarian communities in the world’s Blue Zones.2
Recent studies in support of the Human Microbiome’s role in neurodegenerative disease.
A recent study suggests that the gut microbiome of individuals with preclinical Alzheimer’s disease (AD) is notably different from that of healthy individuals.3 This discovery may help with early diagnosis and provide new avenues for treatment. Specific gut-type identities are associated with preclinical AD, and incorporating microbiome features into the analysis of AD using machine learning could improve the accuracy of predicting preclinical AD. This could also lead to developing therapies that alter the microbiome and prevent cognitive decline.

The study of the microbiome and its role in neurodegenerative diseases is still in its early stages. Researchers are trying to determine whether the gut influences the brain or vice versa. If the microbiome does contribute to the disease state, it is reasonable to assume that changing the microbiome could change the prognosis. Studying the gut microbiome may also help in identifying gut-derived biomarkers for AD. Collecting a stool sample is less invasive and time-consuming than other methods, such as spinal taps or MRI scans.
Various pieces of evidence suggest a role for the gut microbiome in the evolution of AD pathogenesis. During the early stages of AD, clumps of amyloid beta and tau proteins accumulate in the brain.4 Typically, there is a decade-long gap between the first deposition of amyloid beta plaques and the first clinical signs of impairment. Detecting molecular hallmarks of AD early is crucial for effective therapy. Gut microbiome features could enhance early screening measures to identify candidates for follow-up tests such as CSF or PET amyloid beta protein assays.
The study’s findings by Dantas et al. support the hypothesis of an enteric neuroimmune axis in neurodegenerative diseases. More research is needed to determine whether the microbiome of individuals with symptomatic AD differs from those with preclinical AD.5
The microbiota is a significant contributor to various diseases. Studies have demonstrated that transplanting the gut microbiota from ASD patients into germ-free mice resulted in the mice exhibiting autistic behaviors. This is due to the production of neuroactive molecules by the microbiota, which highlights the crucial role of the gut-brain axis in ASD. Microbial metabolites can promote healthy or disease states by activating or deactivating signaling pathways. Altering the gut microbiome can effectively treat or prevent certain diseases. A diet can improve health and longevity by encouraging the growth of healthy microbes and preventing harmful microbial metabolites. Microorganisms can be engineered to treat diseases, and gut-targeted therapies can be developed to reverse microbiome dysbiosis. Engineered probiotics are a viable solution. Machine learning can be used to identify biomarkers and design gut microbe-targeted therapies. It can also assist in creating personalized treatment plans based on individual biological complexity. Analyzing the gut microbiota post-treatment can help predict treatment efficacy.
The role of ML in studying the microbiome.
Machine learning has the remarkable ability to process massive quantities of data. The gut microbiome is highly individualized and dynamic and generates a significant amount of molecular profiling data, such as metabolomics, metagenomics, and metatranscriptomics. Machine learning algorithms have proven invaluable in identifying critical molecular signatures and creating highly accurate models to predict specific phenotypes.
Moreover, machine learning is particularly useful when analyzing the gut microbiome, as it fits perfectly with the P4 medicine model. The role played by dysbiosis of the human microbiome in various disease states, such as diabetes mellitus, inflammatory bowel disease, colorectal cancer, and neurodegenerative disorders, has been well-established. Numerous factors, including genetics, gender, activity level, medication, and diet, influence the complex symbiotic relationship between the gut microbiome and metabolism.
High-throughput technologies are widely used to generate multi-omics data from various human tissues, providing valuable insights into the connections between the gut microbiota and the host. Machine learning is an exceptional tool that explores and integrates multi-omics data, discovers hidden patterns, and generates highly accurate models that predict disease phenotypes. Furthermore, interpretive models can identify potential biomarkers and propose potential therapeutics accurately and confidently.
In the future, machine learning will play an essential role in identifying novel microbial genomes and proteins from uncultured species using gut metagenomic data. This will lead to a better understanding of the microbiota’s mechanisms and functions. Machine learning will also help predict protein structure for drug development and design healthier diets for individuals. It will assist in recommending optimal therapeutics for patients with specific diseases and even play a significant role in developing probiotics to create the ideal microbiome.
Probiotics and prebiotics are microbiome-related therapeutic strategies that have proven beneficial.
Probiotics contain live beneficial bacteria, usually formulated with Lactobacillus and Bifidobacterium, and designed to restore dysbiosis and reduce disease progression and severity. Prebiotics are a substrate that the human body cannot digest, but the bacteria digest them, leading to a health benefit. Prebiotics can be thought of as fuel for beneficial gut bacteria. Studies show the benefits of probiotics/prebiotics in mice models, although the benefits may vary depending on several factors, including formulation and dose.
Faecal Matter Transplant (FMT) has demonstrated efficacy in AD treatment in certain mouse studies, although some studies have been contradictory. Extensive studies are needed to study FMT’s role in AD treatment comprehensively.
Engineered probiotics genetically manipulate bacterial species to produce beneficial metabolites in response to a particular stimulus. This upcoming group of therapeutics has already been used in diseases like Clostridium difficile and inflammatory bowel disease. Targeting inflammatory pathways in the brain regulated by the GMB can influence amyloid deposition.
The Potential of Machine Learning in Neurodegenerative Disease Research
While research into the human microbiome shows promise for understanding and treating neurodegenerative diseases, machine learning, and AI tools also have immense potential to accelerate discoveries in this field. Researchers are applying these technologies to analyze complex neurological data from sources like neuroimaging, genomics, metabolomics, and transcriptomics. The goal is to uncover patterns and insights that can lead to earlier diagnosis, personalized treatments, and new therapeutic targets.
Neuroimaging
Neuroimaging techniques like MRI and PET scanning generate intricate, multidimensional brain images. Machine learning algorithms can analyze these images to identify subtle changes associated with disease progression. For example, convolutional neural networks were able to classify MRI images as belonging to Alzheimer’s disease or normal patients with over 90% accuracy.6 Deep learning models can also predict the conversion from mild cognitive impairment to Alzheimer’s based on structural MRI several years in advance. This could allow earlier intervention and treatment. Beyond classification, machine learning can extract quantitative imaging biomarkers reflective of disease severity from brain images. Such biomarkers enable more objective and sensitive tracking of progression.
Genomics
Genomics research has identified several genetic risk factors for neurodegenerative diseases using genome-wide association studies. However, each variant confers limited individual risk. Machine learning applied to large genomic datasets enables the creation of polygenic risk scores that combine information from multiple variants to predict an individual’s disease risk better. For example, a machine learning model developed polygenic risk scores that could differentiate Parkinson’s disease patients from controls with over 97% accuracy. Furthermore, deep learning algorithms can learn non-linear gene-gene and gene-environment interactions from the data that underlie disease risk.
Metabolomics
Metabolomics investigates metabolic changes associated with disease by profiling small molecule metabolites in biofluids. Machine learning can analyze this complex, high-dimensional data to identify metabolic signatures that are biomarkers of disease onset and progression. For instance, a pipeline combining NMR spectroscopy and machine learning discovered a panel of urinary metabolites that could discriminate Parkinson’s patients from controls. The identified metabolites provide insights into mechanisms and potential therapeutic targets. Multimodal machine learning models that integrate metabolomics with other omics data and clinical information may yield even more robust metabolic signatures of neurodegeneration.
Transcriptomics
Transcriptomics measures gene expression changes across the entire genome. Machine learning applied to this data can determine disease-associated expression profiles involving coordinated changes across multiple genes. For example, a machine learning model identified a blood gene expression signature that distinguishes Alzheimer’s patients from controls. Such expression signatures may reveal disruptions in biological pathways that drive neurodegeneration. They could also enable minimally invasive molecular diagnostics from blood. Beyond bulk transcriptomics, single-cell transcriptomics provides gene expression information at the resolution of individual cells. Machine learning aids in analyzing this vast, high-dimensional data, including identifying cell types and states relevant to disease.
In summary, machine learning shows immense promise as a tool for mining insights from the wealth of omics data being produced in neurodegeneration research. It can help unravel disease mechanisms, discover diagnostic and prognostic biomarkers, and identify potential drug targets. However, machine learning models require extensive, high-quality, multidomain training data. Multidisciplinary collaborations and data-sharing efforts are critical to fully realizing the potential of machine learning in illuminating the complex biology of neurodegenerative disorders.
Other areas for investigation
Diet
The connection between AD and diet is already established. A high-fat diet and obesity are associated with an adverse prognosis and an increased risk of developing the AD phenotype. Mice consuming high-fat diets had accelerated neuropathology and deposition of Amyloid beta plaques. Evidence suggests that a plant-based Mediterranean diet is protective, as observed by Ballarini et al. Diet is one of the most important factors influencing the GMB. More in-depth study is needed to see if diet can alter AD progression through its effect on the GMB.
Sleep
Sleep disturbances and an altered circadian rhythm are heavily implicated in AD. AD suffers tend to sleep during the day and are awake at night with reduced REM and slow wave cycles, essential for memory and cognition. Sleep deprivation aggravates AD. GMB may affect sleep quality, and sleep quality can affect GMB composition. Voigt et al. found that a group of mice called mutant Clock mice that have disrupted sleep lack diversity in their GMB. Disease states such as OSA, IBD, and insomnia showed a connection to the changes in the HMB and sleep disturbances. FMT improved sleep from healthy patients to patients with IBD. There is likely an essential connection between AD, sleep, and GMB, but it needs further study.
Exercise
Studies have shown that exercise reduces age-related cognitive decline and AD risk. This is related to enhanced adult hippocampal neurogenesis, improved brain-derived neurotrophic factor signaling, and reduced neuroinflammation. Exercise modifies the composition and diversity of the GMB. Allen et al. showed that exercise changed the composition of the GMB. Exercise is considered protective in AD, and further study into its benefits as part of a treatment plan is necessary.
Conclusion
Consuming all of this information may indeed be “food for thought,” keeping in mind that:
“Without memory, there is no culture. Without memory, there would be no civilization, no society, no future.” – Elie Wiesel.
As we look to an increasingly aging society, we have to ask ourselves where we should best invest our academic resources in the search for a “cure-all” rather than a Band-Aid, particularly if, like William Gibson, we lament the fact that: “Time moves in one direction, memory in another.”
The human microbiome and advanced AI/machine learning techniques both show tremendous promise as avenues of research that could unravel the complex biology underlying neurodegenerative diseases. Integrating multiple streams of omics data using AI may provide the critical insights needed to develop preventive microbiome-directed therapies. With a multifaceted approach leveraging human ingenuity alongside increasingly intelligent algorithms, we can move closer to a future where debilitating brain disorders are caught early and their progression halted or reversed. Our memories, relationships, and quality of life depend on it.
References
- Walker, Douglas, et al. “Studies on Colony Stimulating Factor Receptor-1 and Ligands Colony Stimulating Factor-1 and Interleukin-34 in Alzheimer’s Disease Brains and Human Microglia.” Frontiers in Aging Neuroscience, vol. 15, 2017, doi:10.3389/fnagi.2017.00115.
- Palmas, Valeria, et al. “Gut Microbiota Markers and Dietary Habits Associated with Extreme Longevity in Healthy Sardinian Centenarians.” Nutrients, vol. 14, no. 12, 2022, p. 2436. https://doi.org/10.3390/nu14122436.
- Ferreiro, Ana Lia, et al. “Gut microbiome composition may be an indicator of preclinical Alzheimer’s disease.” Science translational medicine vol. 15,700 (2023): eabo2984. doi:10.1126/scitranslmed.abo2984
- “Brain Science.” Gray Connections, https://grayconnections.net/category/brain-science/.
- Dantas, Gautam, et al. “Gut Microbiome Composition May Be an Indicator of Preclinical Alzheimer’s Disease.” Science Translational Medicine, vol. 15, no. 700, 2023, doi:10.1126/scitranslmed.abo2984.
- Korolev, Sergey, et al. “Predicting progression from mild cognitive impairment to Alzheimer’s dementia using clinical, MRI