Recent studies have shown a growing interest in the complex relationship between the human gut microbiota, metabolism, and overall health. This review aims to explore the gut microbiota–host association, focusing on its implications for precision nutrition and personalized medicine. The objective is to highlight how gut microbiota modulate metabolic and immune functions, contributing to disease susceptibility and wellbeing. The review synthesizes recent research findings, analyzing key studies on the influence of gut microbiota on lipid and carbohydrate metabolism, intestinal health, neurobehavioral regulation, and endocrine signaling. Data were drawn from both experimental and clinical trials examining microbiota–host interactions relevant to precision nutrition. Our findings highlight the essential role of gut microbiota-derived metabolites in regulating host metabolism, including lipid and glucose pathways. These metabolites have been found to influence immune responses and gut barrier integrity. Additionally, the microbiota impacts broader physiological processes, including neuroendocrine regulation, which could be crucial for dietary interventions. Therefore, understanding the molecular mechanisms of dietary–microbiota–host interactions is pivotal for advancing personalized nutrition strategies. Tailored dietary recommendations based on individual gut microbiota compositions hold promise for improving health outcomes, potentially revolutionizing future healthcare approaches across diverse populations.

The human gut microbiome, an intricate ecosystem housing trillions of microorganisms within the gastrointestinal tract, holds significant importance in human health and the development of diseases. Recent advances in technology have allowed for an in-depth exploration of the gut microbiome, shedding light on its composition and functions. Of particular interest is the role of diet in shaping the gut microbiome, influencing its diversity, population size, and metabolic functions. Precision nutrition, a personalized approach based on individual characteristics, has shown promise in directly impacting the composition of the gut microbiome. However, to fully understand the long-term effects of specific diets and food components on the gut microbiome and to identify the variations between individuals, longitudinal studies are crucial. Additionally, precise methods for collecting dietary data, alongside the application of machine learning techniques, hold immense potential in comprehending the gut microbiome’s response to diet and providing tailored lifestyle recommendations. In this study, we investigated the complex mechanisms that govern the diverse impacts of nutrients and specific foods on the equilibrium and functioning of the individual gut microbiome of seven volunteers (four females and three males) with an average age of 40.9 ± 10.3 years, aiming at identifying potential therapeutic targets, thus making valuable contributions to the field of personalized nutrition. These findings have the potential to revolutionize the development of highly effective strategies that are tailored to individual requirements for the management and treatment of various diseases.

Nutrition is a cross-cutting sector in medicine, with a huge impact on health, from cardiovascular disease to cancer. Employment of digital medicine in nutrition relies on digital twins: digital replicas of human physiology representing an emergent solution for prevention and treatment of many diseases. In this context, we have already developed a data-driven model of metabolism, called a “Personalized Metabolic Avatar” (PMA), using gated recurrent unit (GRU) neural networks for weight forecasting. However, putting a digital twin into production to make it available for users is a difficult task that as important as model building. Among the principal issues, changes to data sources, models and hyperparameters introduce room for error and overfitting and can lead to abrupt variations in computational time. In this study, we selected the best strategy for deployment in terms of predictive performance and computational time. Several models, such as the Transformer model, recursive neural networks (GRUs and long short-term memory networks) and the statistical SARIMAX model were tested on ten users. PMAs based on GRUs and LSTM showed optimal and stable predictive performances, with the lowest root mean squared errors (0.38 ± 0.16–0.39 ± 0.18) and acceptable computational times of the retraining phase (12.7 ± 1.42 s–13.5 ± 3.60 s) for a production environment. While the Transformer model did not bring a substantial improvement over RNNs in term of predictive performance, it increased the computational time for both forecasting and retraining by 40%. The SARIMAX model showed the worst performance in term of predictive performance, though it had the best computational time. For all the models considered, the extent of the data source was a negligible factor, and a threshold was established for the number of time points needed for a successful prediction.

VO₂max index has a significant impact on overall health. Its estimation through wearables notifies the user of his level of fitness but cannot provide a detailed analysis of the time intervals in which heartbeat dynamics are changed and/or fatigue is emerging. Here, we developed a multiple modality biosignal processing method to investigate running sessions to characterize in real time heartbeat dynamics in response to external energy demand. We isolated dynamic regimes whose fraction increases with the VO₂max and with the emergence of neuromuscular fatigue. This analysis can be extremely valuable by providing personalized feedback about the user’s fitness level improvement that can be realized by developing personalized exercise plans aimed to target a contextual increase in the dynamic regime fraction related to VO₂max increase, at the expense of the dynamic regime fraction related to the emergence of fatigue. These strategies can ultimately result in the reduction in cardiovascular risk.

Development of predictive computational models of metabolism through mechanistic models is complex and resource demanding, and their personalization remains challenging. Data-driven models of human metabolism would constitute a reliable, fast, and continuously updating model for predictive analytics. Wearable devices, such as smart bands and impedance balances, allow the real time and remote monitoring of physiological parameters, providing for a flux of data carrying information on user metabolism. Here, we developed a data-driven model of end-user metabolism, the Personalized Metabolic Avatar (PMA), to estimate its personalized reactions to diets. PMA consists of a gated recurrent unit (GRU) deep learning model trained to forecast personalized weight variations according to macronutrient composition and daily energy balance. The model can perform simulations and evaluation of diet plans, allowing the definition of tailored goals for achieving ideal weight. This approach can provide the correct clues to empower citizens with scientific knowledge, augmenting their self-awareness with the aim to achieve long-lasting results in pursuing a healthy lifestyle.

Self-monitoring of weight, diet and physical activity is a valuable component of behavioral weight loss treatment. The validation and user-friendliness of this approach is not optimal since users are selected from homogeneous pools and rely on different applications, increasing the burden and achieving partial, generic and/or unrelated information about their metabolic state. Moreover, studies establishing type, time, duration, and adherence criteria for self-monitoring are lacking. In this study, we developed a digital web-based application (ArmOnIA), which integrates dietary, anthropometric, and physical activity data and provides a personalized estimation of energy balance. Moreover, we determined type, time, duration, and adherence criteria for self-monitoring to achieve significant weight loss in a highly heterogeneous group. A single-arm, uncontrolled prospective study on self-monitored voluntary adults for 7 months was performed. Hierarchical clustering of adherence parameters yielded three behavioral approaches: high (HA), low (LA), and medium (MA) adherence. Average BMI decrease is statistically significant between LA and HA. Moreover, we defined thresholds for the minimum frequencies and duration of dietary and weight self-monitoring. This approach can provide the correct clues to empower citizens with scientific knowledge, augmenting their self-awareness with the aim of achieving long-lasting results when pursuing a healthy lifestyle.