Slimness and longevity algorithms: how mathematical models and artificial intelligence create your ideal nutrition plan

Біговий захід GoMove в Ужгороді за підтримки Ediens

ABSTRACT: The era of universal diets, designed only for counting calories, is coming to an end. Modern science proves: our body's reaction to food is completely individual and 100% depends on the trillions of microorganisms that inhabit our intestines. Today, advanced 4P medicine (predictive, preventive, personalized, participatory) combines deep genetic research of the microbiome with powerful IT algorithms and machine learning methods. In this article, we will analyze in detail how a personalized nutrition plan is created, why a regular ChatGPT will never replace clinical databases, and how the digitalization of medicine allows us to use food as the most powerful medicine.

PART I. The End of the Era of Universal Diets: The One Apple Paradox

For decades, nutrition was based on general rules: "eat less fat," "count calories," "here's a universal diet for weight loss." However, in practice, doctors and patients have encountered the fact that the same diet causes weight loss and energy in one person, and bloating, fatigue, and exacerbation of chronic diseases in another.

Why is this happening? The answer lies in the realm of microbiomics. A large-scale cohort study (published by Zeevi et al. in 2015) turned the idea of metabolism upside down: scientists found an extremely high variability in the response to identical foods in different people. If two different people are given the same apple, their glycemic response (blood sugar spike) and the amount of energy absorbed will be radically different.

The secret is that when we eat, we are not so much feeding ourselves as we are feeding our own microorganisms. It is this unique composition of bacteria that determines how nutrients are broken down, which metabolites (such as beneficial short-chain fatty acids) enter the bloodstream, and whether this food is converted into energy or fat. Therefore, true therapeutic nutrition cannot be universal - it must be based on the exact data of your body.

PART II. The Data Triad: What "feeds" a digital algorithm?

In order to create a truly working, medically sound nutrition plan , the system needs a huge array of input data. Ediens’ approach is based on collecting three key components:

Gut microbiome genetic analysis (NGS - next generation sequencing): This is the foundation. The analysis determines the DNA of all microorganisms, reveals the person's enterotype, species balance (Shannon and Pillow indices) and, most importantly, provides metabolomic analysis. The system understands not only wholives in the gut, but alsowhat they do: what toxins or beneficial substances they produce.
Biochemical blood test: Blood reflects the body's current physiological response to the state of the microbiome and any existing deficiencies or systemic inflammation.
Deep Digital History (Quiz/Questionnaire) : This is a database of the patient’s phenotype. The algorithm takes into account physical parameters (height, weight), existing diagnosed diseases, lifestyle, stress level, place of residence (which affects the availability of local products), and eating habits.“If a person has never eaten avocado, the algorithm will not force them to do so, but will find an alternative that performs the same function.” The success of a diet depends on 50% on the needs of the microbiome, and 50% on the preferences of the person himself.

PART III. Algorithms vs. Artificial Intelligence: Why ChatGPT Can't Be Your Nutritionist

With the rise of neural networks, it’s tempting to simply upload your tests to an open chatbot and get a diet. However, in evidence-based medicine, this approach is considered dangerous.

Open large language models (LLMs, such as the basic ChatGPT) are prone to “hallucinations” — they can invent non-existent scientific articles or generate inadequate recommendations that do not take into account the complex limiting factors of human health. For example, an open AI may recommend eating a lot of cabbage to stimulate certain bacteria, completely ignoring the fact that the patient has bacterial overgrowth syndrome (SIBO) or gallbladder problems, in which this product will cause a sharp deterioration in the condition.

That is why cutting-edge science uses mathematical and predictive models . The Ediens information system is based on multi-year cohort studies and uses the principle of principal components analysis (PCA) . How does this work at the machine learning level?

The system operates with its own and international closed databases (e.g. European Food Indexation Resource - EuroFIR, ePlantLIBRA). It knows the exact micro- and macroelement composition of hundreds of products.
The algorithm contains a table called "Plant_CM (gut microbiome)" , which calculates how a specific biologically active substance from a vegetable or fruit activates or inhibits (suppresses) a specific strain of bacteria in your gut.
Next, the "disease sieve" kicks in. If the algorithm sees that turmeric is a perfect fit for your microbiome, but your questionnaire indicates an allergy or gastritis in the acute stage, the system mathematically rejects this product and looks for a safe analogue that will perform the same biochemical task.

This is not just text generation, it is solving complex multi-criteria mathematical problems to achieve target homeostasis.

PART IV. From Math to Plate: What Does the Final Plan Look Like?

After the algorithm has processed gigabytes of genetic and clinical data, it moves on to practical implementation — diet formation.

Calculating energy needs: Using the Mifflin-Saint-Geor formula, the ideal calorie intake is calculated taking into account the patient's target weight.
Macronutrient balance: There is a division into proteins, fats, and carbohydrates (most often according to a low-glycemic protocol close to metabolic balance, for example, in a ratio of 20:35:45).
Targeted Menu (14-Day Cycle): Unlike conventional weekly diets, the algorithm generates a 14-day plan. This is the optimal time for the plant ingredients (which act as prebiotics) to smoothly modulate the composition of the intestinal microbiota without causing sudden stress.
Meal Selection: The plan doesn't just contain a dry list of "eat more fiber." The system provides lists of recommended and categorically not recommended (at the moment) foods, and also creates a daily meal plan (breakfast, lunch, dinner) with a specific number of grams of each component.

Special emphasis is placed on local and traditional fermented foods, as the microbial starters in them (probiotic agents) are best absorbed by the body in a specific geographical area.

CONCLUSION

Artificial intelligence and mathematical algorithms in nutritional science are not science fiction, but the only tool capable of processing the astronomical number of connections between bacterial genes, our blood, and food components. The personalized nutrition plan resulting from such a symbiosis ceases to be just a "diet for weight loss." It turns into your personal medical protocol - a lifestyle that purposefully relieves inflammation, restores energy, prevents serious non-communicable diseases, and programs the body for a healthy and active longevity.

❓ CLINICAL Q&A: 4 questions about algorithmic nutrition

1. Why can’t I just upload my tests to the free ChatGPT and get a diet? Open language models (LLMs) generate text based on word probabilities, not clinical protocols. They often “hallucinate” and can recommend products that contradict your comorbidities (e.g., prescribe aggressive fiber for acute intestinal inflammation). A true medical IT algorithm is based on rigorous mathematical models, the principle of principal components (PCA), and closed verified databases of product composition (e.g., EuroFIR).

2. If the algorithm calculates that a product that I can’t stand is good for me, will I have to force myself to eat it? No. Before running the algorithm, you fill out a detailed questionnaire (Quiz) where you indicate your eating habits, intolerances, and products that you absolutely do not like. The algorithm has hundreds of replacement options: if a conventional avocado doesn’t suit you, the system will mathematically select another product or spice with an identical composition of trace elements and polyphenols that will perform the same therapeutic function for your bacteria.

3. What is the period for which such an algorithmic nutrition plan is designed? Usually, the core of the plan is designed for a 14-day period. This cycle is physiologically optimal in order to adequately distribute the necessary biologically active substances and smoothly adjust the bacterial population without stressing the digestive system. After passing the therapeutic or corrective stage, these nutritional principles adapt and become your permanent intuitive lifestyle.

4. Can this diet plan cure chronic diseases such as diabetes or obesity? Food itself is not a direct pill for the disease, but it is the most powerful tool for epigenetic regulation. Personalized nutrition does not "cure" symptoms, it eliminates the root cause - systemic inflammation and metabolic imbalance. By correcting the microbiome with the help of properly selected prebiotics from food, we restore normal neurotransmitter synthesis, improve cell sensitivity to insulin and stop the development of atherosclerosis or type II diabetes at the molecular level