After 12 weeks of following a VLCD (Phase1), participants begin a six-week transition (Phase 2 ). During this phase, they adopt a CH-reduced diet, consuming 25-30% of their total energy from carbohydrates. This involves shifting gradually but structurally from formula products to regular meals. Meal boxes and formula products aligned with their assigned diet support this shift and serve as educational tools. Phase 2 also includes an exercise program consisting of two supervised 1-hour sessions and one 1-hour unsupervised high-intensity session weekly (intensity >70% peak oxygen uptake (VO2peak) and/or >7 on the RPE scale, equivalent to 1-3 repetitions in reserve for resistance training or vigorous intensity). For the next 34 weeks (Phase 3), participants receive ongoing diet and exercise support while purchasing and preparing their own meals according to their assigned diet. They continue with two supervised and one unsupervised group session peer week.

The degree of implementation of intervention activities and the extent of participant engagement will be measured by tracking attendance and completion of planned intervention components, including intervention visits, supervised and unsupervised exercise sessions, phone calls, at-home measurements, group-based education sessions, and peer support activities. Data will be collected via REDCap (8.10.18, Vanderbilt University, TN, USA) and study logs to quantify the proportion of completed versus scheduled activities for each participant.

Group-based diet and exercise sessions will be observed to assess how the intervention is delivered in practice and how participants engage with the activities. Field notes will document group dynamics, adherence to session protocols, and contextual factors. Field notes will also inform interview guides and supplement interview data.

Total number of days during which participants did not adhere to the planned intervention protocol or experienced protocol deviations during the three intervention phases.

Interviews with participants, dieticians and exercise instructors.

Dietary intake, including energy intake (kcal/day) and macronutrient intake (total in grams and E% per day) and micronutrient intake (units/day) based on 3-days' dietary recalls based on MyFood24.

Number of participants with HbA1c < 48 mmol/mol maintained for at least 3 months without glucose-lowering medication.

Measured to nearest 0.1 kg using a digital scale. Fasted state.

Number of participants who achieved a weight loss of at least 10% of their baseline body weight. Measured to nearest 0.1 kg using a digital scale. Fasted state.

Assessed from blood samples collected in non-fasting state.

Assessed from blood samples collected in non-fasting state.

Estimated glomerular filtration rate (eGFR).

Concentration of creatinine, assessed from blood sample in fasting state.

Concentration of potassium, assessed from blood sample in fasting state.

Concentration of sodium, assessed from blood sample in fasting state.

Fasting concentration of plasma albumin.

Fasting concentration of plasma hemoglobin.

Fasting concentration of white blood cells.

Measured using a using an ultrasonic device to nearest 0.1 cm in order to calculate BMI.

Measured to nearest 0.1 kg, in the morning, after urinating, before ad libitum diet acc to ad libitum intake of allocated intervention diet, using wireless scales at participants' home.

Calculated from body weight (kg) and height (m). Fasted state.

Fecal samples will be analyzed for microbiome composition and diversity using 16S rRNA gene sequencing and shotgun metagenomic sequencing.

Participants will collect three consecutive morning midstream urine samples and four additional spot urine samples at home. Urine samples will be analyzed by metabolomic profiling to quantify urinary concentrations of at least 10 dietary biomarkers, including Sulforaphane-N-acetylcysteine (broccoli), Proline betaine (citrus fruits), Carnosine (meat), and 3,5-dihydroxybenzoic acid glucuronides (whole grains).

Urine spot samples will be collected to assess renal function and diabetes-related kidney involvement.

Saliva samples, analyzed for microbial composition changes via 16S rRNA and metagenomic.

Concentrations of metabolites including, but not limited to, glucose, triglycerides, free fatty acids (FFA), and alanine will be assessed during the MMTT at 8 time points; before the meal while fasting (minutes -15 and -5) and postprandially (minutes 30, 60, 90, 120, 180, and 240). Additionally, concentrations of lipoproteins and intact and split proinsulin will be measured in the fasting samples.

Optionally, concentrations of hormones involved in glucose and lipid metabolism and appetite regulation including insulin, glucagon, C-peptide, GLP-1, GIP, ghrelin, PYY, CCK and more variables will be assessed during the MMTT at 8 time points; before the meal while fasting (minutes -15 and -5) and postprandially (minutes 30, 60, 90, 120, 180, and 240). Results will be expressed in pmol/L. Area under the curve (AUC) will be expressed as pmol/L*min.

Insulin sensitivity will be estimated from postprandial responses of relevant metabolites obtained during the MMTT using validated model-based indices.

β-cell function will be estimated from postprandial responses of relevant metabolites obtained during the MMTT using validated model-based indices.

Hepatic first-pass insulin extraction will be estimated from postprandial responses of relevant metabolites obtained during the MMTT using validated model-based indices.

Metabolic insulin clearance will be estimated from postprandial responses of relevant metabolites obtained during the MMTT using validated model-based indices.

Rated using visual analogue scales (VAS) and includes sensations of: Hunger, fullness, satiety, prospective food consumption, wellbeing, nausea, thirst, desire to eat meat, salty, or sweet. The scale range is 0-100 and each end represent the extremes e.g. hunger rating: "I am not hungry at all" to "I have never been this hungry before". VAS ratings are collected at 7 time points during the MMTT; before the meal while fasting (minute -10) and postprandially (minutes 30, 60, 90, 120, 180, 240).

Physical activity (PA) including but limited to: PA (time spent at different intensities (sedentary activity, low light activity, high light activity, moderate to vigorous activity, vigorous activity), PA (total counts, counts/min), PA energy expenditure (kcal/day), PA (MET hours) and timing of PA (hh:mm) will be objectively assessed using a thigh-mounted accelerometer (SENS Motion). Participants will wear the device 24 hours/day for 10 consecutive days prior to visits. Data will provide time-stamped estimates of sedentary time, time spent standing, walking, bicycling, and intensity of movement.

Measured using tape measure to the nearest 0.5 cm. Fasted state.

Measured using tape measure to the nearest 0.5 cm. Fasted state.

The ratio of the circumference of the waist to that of the hips.

Measured by Dual-energy X-ray Absorptiometry in a fasted state.

Measured by Dual-energy X-ray Absorptiometry in a fasted state.

Measured by Dual-energy X-ray Absorptiometry in a fasted state.

Measured by Dual-energy X-ray Absorptiometry in a fasted state.

Measured by FibroScan in a fasted state.

Measured by FibroScan in a fasted state.

Will be calculated from participants' age and fasting concentration of aspartate-aminotransferase (U/L), alanine-aminotransferase (U/L) and thrombocytes (x10^9/L).

Cognitive functions will be assessed by Screen for Cognitive Impairment in Psychiatry Danish Version (SCIP-D) together with the Trail Making Test Part B (TMT-B, psychomotor speed and executive function).

Food choice of food items from four combined food categories (high-fat savoury, high-fat sweet, low-fat savoury and low-fat sweet foods) examined from the Steno Biometric Food Preference Task (SBFPT). Food choice is determined based on frequency of selection made within each food category. The scores range from 0-48 i.e. 0 = foods within a specific food category have not been selected at all to 48 = foods within a specific food category have been selected 48 times.

Measured using eye tracking in response to looking at food pictures during Steno Biometric Food Preference Task (SBFPT). Includes the following parameters: Gaze: Time spent (ms and %) and visits (n); and fixations: Time to first fixation (ms), time spent (ms and %), fixation count (n), first fixation duration (ms), average fixation duration (ms). Distance to screen (mm), and gaze direction bias (ratio) which is calculated as the number of trials in which the first fixation was directed to a food image as a proportion to all trials. A bias score >0.5 indicates attention towards one food image, a bias score equal to 0.5 indicates no bias, and a bias score <0.5 indicates attention towards the other food images.

Reaction time during forced food choice of food items from four combined food categories (high-fat savoury, high-fat sweet, low-fatsavoury and low-fat sweet foods) examined from the Steno Biometric Food Preference Task (SBFPT).

Explicit liking of 16 food items from four combined food categories (high-fat savoury, high-fat sweet, low-fat savouryand low-fat sweet foods) examined from the Steno Biometric Food Preference Task (SBFPT). Explicit liking is rated using visual analogue scales and the range is 0-100. Each end represents the extremes e.g. Question: "how pleasant would it be to taste this foodright now?" Answer: "not at all" (rated 0 on the 0-100 scale) to "extremely" (rated 100 on the 0-100 scale).

Implicit wanting of food items from four combined food categories (high-fat savoury, high-fat sweet, low-fat savoury and low-fat sweet foods) examined from the Steno Biometric Food Preference Task (SBFPT). Implicit wanting is assessed based on food choice and response time for selected and non-selected food items as well as mean response time (a frequency-weighted algorithm). In this frequency-weighted algorithm a positive score indicates a morerapid preference for a food type over another food type and a negative score indicates the opposite. A score of zero indicates that food types are equally preferred. The frequency weighted algorithm isused so the implicit wanting score is influenced by both selection (positively contributing to the score) and non-selection (negatively contributing to the score) of food type. Scores for implicit wanting typically range from -100-100 (due to reaction time there is no fixedmin-max value).

Explicit wanting of 16 food items from four combined food categories (high-fat savoury, high-fat sweet, low-fat savoury and low-fat sweet foods) examined from the Steno Biometric Food Preference Task (SBFPT). Explicit wanting is rated using visual analogue scales and the range is 0-100. Each end represents the extremes e.g. Question: "how much do you want some of this food now?" Answer: "not at all" (rated 0 on the 0-100 scale) to "extremely" (rated 100 on the 0-100 scale).

Unspecified exploratory outcomes related to the 16 food items from four combined food categories (high-fat savoury, high-fat sweet, low-fat savoury and low-fat sweet foods) examined from the Steno Biometric Food Preference Task (SBFPT).

Measured under resting and fasting conditions.

Measured under resting and fasting conditions.

Measured under resting and fasting conditions.

A standard 12-lead resting ECG will be performed to assess cardiac rhythm and detect potential contraindications for maximal exercise testing. This screening is conducted to ensure participant safety prior to performing the VO₂ peak test.

Peak oxygen consumption (VO2peak) as a measure of cardiorespiratory fitness will be measured by an incremental bicycle ergometer test.

Peak oxygen consumption (VO2peak) per kg body weight as a measure of cardiorespiratory fitness. Measured by an incremental bicycle ergometer test.

Measured using a blinded continuous glucose monitoring device (CGM). The sensor will be placed subcutaneously on the upper arm or abdomen, and worn for 10-14 days prior to visits.

Measured using a blinded continuous glucose monitoring device (CGM). The sensor will be placed subcutaneously on the upper arm or abdomen, and worn for 10-14 days prior to visits.

Measured using a blinded continuous glucose monitoring device (CGM). The sensor will be placed subcutaneously on the upper arm or abdomen, and worn for 10-14 days prior to visits.

Measured using a blinded continuous glucose monitoring device (CGM). The sensor will be placed subcutaneously on the upper arm or abdomen, and worn for 10-14 days prior to visits.

Assessed from GPS-based smart watch.

Assessed using the Quality of Life (12-Item Short Form Survey, SF-12), which measures health-related quality of life across physical and mental health domains. The questionnaire includes 12 items and yields two summary scores: the Physical Component Summary (PCS) and the Mental Component Summary (MCS). Scores range from 0 to 100, with higher scores indicating better health status. A PCS score of 50 or below may indicate physical health impairment, while an MCS score of 42 or below may suggest clinical depression.

Assessed from the questionnaire European Quality of Life - 5 Dimensions (EQ-5D). EQ-5D comprises 5 dimensions (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression), each rated on 3 or 5 levels of severity. A health utility index score is derived, ranging from values below 0 (where 0 represents death and negative values represent states worse than death) to 1 (full health). Higher scores indicate better health-related quality of life.

Measure of severity of symptom burden. The number of non-missing (answered) items in the 25-item Bodily Distress Syndromes (BDS) checklist (from bds1 to bds25) is counted for each individual to determine if enough data is available to compute a score. Then, if at least 13 items are answered, the average of the available responses is calculated, scaled to a 0-100 range, rounded to the nearest whole number, and saved as the BDS sum score. Lower value is better.

For each of the following domains: cardiopulmonary symptoms (b1-b6), gastrointestinal symptoms (b7-b13), musculoskeletal symptoms (b14-b20) and general symptoms (b21-b25).Bimodal variable for cases - Number of symptoms scored above 1 (including Somewhat, Quite a bit & A lot). Participants reporting ≥4 such symptoms and with fewer than 3 missing responses will be classified as a case being at least four symptoms within each domain (binary variable: 1 = case, 0 = non-case).

Assessed from the Problem Areas in Diabetes Scale (PAID-5 scale) comprising five of the emotional-distress questions of the full PAID items. Each item can be rated from 0 to 4. A total score of ≥8 indicates possible diabetes related emotional distress.

Assessed from the questionnaire The Three Item Loneliness Scale (TILS). TILS consists of 3 items assessing subjective feelings of loneliness and social isolation. Scores range from 3 to 9, with higher scores indicating greater loneliness. A score of 6 or above has been used as a cut-off to indicate elevated loneliness in population-based studies.

Assessed from the questionnaire Pittsburgh Sleep Quality Index. The questionnaire consists of 19 items. Each item is weighted on a 0-3 interval scale. The global PSQI score is then calculated by totaling the seven component scores, providing an overall score ranging from 0 to 21, where lower scores denote a healthier sleep quality.

Assessed from the questionnaire Self-efficacy for nutrition change (SEFNC). SEFNC assesses individuals' confidence in their ability to maintain healthy eating habits in challenging situations over the next six months. The scale includes 9 items addressing common barriers such as emotional distress, lack of support, time pressure, and disrupted routines. Each item is rated on a percentage scale from 0% ("Not at all confident") to 100% ("Completely confident"), in 10% increments. A higher mean score indicates greater self-efficacy for sustaining nutrition-related behavior change.

Assessed from the questionnaire Social Support for Eating Habits Questionnaire. The questionnaire includes 10 items rated separately for family and friends, assessing both supportive and unsupportive behaviors related to dietary change. Items are rated on a scale from 1 (Not at all) to 5 (Very often), with 8 indicating "Not applicable." Higher scores on supportive items reflect greater perceived support; higher scores on unsupportive items reflect greater social barriers.

Assessed from the questionnaire Yale food addiction scale (YFAS). It consists of 25 items evaluating symptoms such as loss of control over eating, continued use despite negative consequences, and withdrawal. The scale provides both a symptom count score (ranging from 0 to 11) and a diagnostic threshold indicating the presence of food addiction. Higher scores indicate more severe food addiction symptomatology. A diagnosis is assigned if a certain number of symptoms are endorsed along with clinically significant impairment or distress

Number and description of adverse events and serious adverse events reported during the intervention period.

Heart rate assessed from GPS-based smart watch and corresponding %VO2peak (measured at test visits).

Rate of perceived Exertion on a 0 to 10 scale (0 being no exertion at all and 10 being a maximum effort).

Cardiac output (L/min) will be measured during the MMTT for 5 min using the non-invasive Finapres Nova device at minute -30 (pre-meal baseline), and at minutes 15, 45, 75, 105, 135, 195, 255.

Systemic vascular resistance (dyn*s/cm5) will be measured during the MMTT for 5 min using the non-invasive Finapres Nova device at minute -30 (pre-meal baseline), and at minutes 15, 45, 75, 105, 135, 195, 255.

Heart rate (bpm) and heart rate variability (ms) will be measured during the MMTT for 5 min using the non-invasive Finapres Nova device at minute -30 (pre-meal baseline), and at minutes 15, 45, 75, 105, 135, 195, 255.

Blood pressure and blood pressure variability (mmHg) will be measured during the MMTT for 5 min using the non-invasive Finapres Nova device at minute -30 (pre-meal baseline), and at minutes 15, 45, 75, 105, 135, 195, 255.

Sleep timing (hh:mm) will be measured with a GPS-based smartwatch and a thigh-mounted accelerometer (SENS Motion).

Sleep duration (minutes) will be measured with a GPS-based smartwatch and a thigh-mounted accelerometer (SENS Motion).

Sleep variability (minutes) will be measured with a GPS-based smartwatch and a thigh-mounted accelerometer (SENS Motion).

Sleep efficiency (%) will be measured with a GPS-based smartwatch and a thigh-mounted accelerometer (SENS Motion).

Sleep wakefulness (%) will be measured with a GPS-based smartwatch and a thigh-mounted accelerometer (SENS Motion).

Systolic and diastolic blood pressure (mmHg) will be measured at home under resting conditions.

Resting heart rate (bpm) will be measured at home under resting conditions

Capillary blood glucose (mmol/L) will be measured at home before the three main meals and at bedtime over two consecutive days.