On the basis of our primary study outcome (feasibility), we will conduct a 2-phase single-arm study [42,43] to preliminarily test our mHealth-based fiber intervention in up to 80 young adults with prediabetes (phase 1) and elucidate factors that impact sustained behavior change and fiber intake after intervention end using EMAs (phase 2) (Figure 1).

The theoretical framework for this intervention is a combination of self-determination theory (SDT), social cognitive theory (SCT), and the biopsychosocial model [44-46]. In brief, the intervention taps into SDT given that it honors individual autonomy (eg, allows for flexibility in the remainder of the diet outside of fiber), promotes competence (eg, targets knowledge), and promotes relatedness (eg, use of interactive mHealth content). It also taps into SCT given that it alters cognitive processes and integrates suggestions from previous participants (observational learning). Finally, it intentionally integrates elements of coping to address the expected yet physiologically normal GI responses that can occur with increased fiber ingestion [21,22]. These messages are informed by both cognitive behavioral therapy and acceptance and commitment therapy, featuring elements of anticipatory coping (eg, preparing individuals for potential symptoms), behavioral coping strategies (eg, gradual fiber increases, hydration, and symptom monitoring), and elements of cognitive reframing [47].

We will recruit young adults aged 18 to 39 years who have prediabetes [48]. Classification of prediabetes is based on having a hemoglobin A_1c (HbA_1c) level between 5.7% and 6.4% per American Diabetes Association guidelines [49]. Other inclusion criteria are residing in or near Tulsa, Oklahoma (where in-person data collection will take place), or nearby and being willing to travel for visits; ability to access and/or use a smartphone; and proficiency in the English language. Exclusion criteria are a suspected eating disorder based on the 26-item Eating Attitudes Test (which was reproduced with permission [50]), current use of a glucagon-like peptide-1 receptor agonist, food allergies or intolerances that preclude engagement, self-reported diagnosis of irritable bowel syndrome or inflammatory bowel disease, self-reported pregnancy or breastfeeding, or travel plans that limit intervention completion.

Participants will be recruited via community events; email distributions to students, faculty, and staff at the University of Oklahoma; directly via the medical record system of the University of Oklahoma Health Campus (MyChart); social media advertisements; provision of flyers at university health screenings; and direct communication and distribution of flyers in the clinics within internal medicine at the University of Oklahoma Health Campus in Tulsa. Prospective participants can provide documentation of an HbA_1c level within the prediabetes range from the previous 3 months or complete the Centers for Disease Control and Prevention Prediabetes Risk Test as part of the study screening. If prospective participants are otherwise eligible and appear to be at risk of prediabetes per the Prediabetes Risk Test, they will be invited to attend an in-person recruitment visit wherein they can complete the informed consent process, followed by a screening HbA_1c test. If their HbA_1c level is within the prediabetes range, they will be eligible for enrollment in the study. If their HbA_1c level is not within the prediabetes range, their participation will be considered complete, and they will be deemed ineligible.

University of Oklahoma Health Campus Institutional Review Board approval was obtained on August 18, 2025. Written informed consent will be obtained from all individuals in person prior to any study-related screening of HbA_1c and (if eligible) study participation. During the informed consent process, participants will be informed of potential risks, including those that are expected to occur with increases in fiber intake (eg, GI discomfort). Participants will receive compensation in the form of gift cards for completion of initial HbA_1c screening (US $15), US $50 for the follow-up visit if applicable, and up to US $168 for EMAs across both phases (up to US $84 for phase 1 and up to US $84 for phase 2). Rates of EMA compensation will be based on the rate of EMA completion: participants who complete 50% (42/84) to 73.8% (62/84) of EMAs will receive a US $42 gift card, those who complete 75% (63/84) to 89.3% (75/84) of EMAs will receive a US $63 gift card, and those who complete 90.2% (>75/84) or more will receive a US $84 gift card [51].

Data will be primarily collected in REDCap (Research Electronic Data Capture; Vanderbilt University). EMA data will be collected in the University of Oklahoma Health Campus OU Health Stephenson Cancer Center mHealth Shared Resource (Insight). No protected health information will be stored or transmitted via Insight. CGM data will be accessed via participant Clarity app or Dexcom accounts for data extraction. All data will be ultimately coded and saved on OU Health specific platforms and/or REDCap. EMA and CGM data will be monitored for the corresponding intervention periods. If blood glucose levels remain abnormally elevated, consultation with the remainder of the team (eg, MJJ and LJC) will be made, and appropriate action will be taken. Participants will also be given guidance on actions to take if blood glucose remains high. In addition to study team contact information, participants will be provided with an individualized REDCap link to report any adverse events, which will be promptly reviewed by the research team and reported according to (1) whether they were related to the study, (2) whether they were anticipated, and (3) level of severity. Participants will be told that, if any foods or procedures cause discomfort, they are free to refrain from eating those foods or completing those procedures without compromising total participation in the study.

This protocol describes a first-of-its-kind examination of a fully remote mHealth intervention that focuses singularly on promoting intake of fiber in young adults with prediabetes. This protocol and resulting data will shed light on two key factors: (1) the utility of an mHealth intervention in young adults and (2) the potential effects of a singular focus on fiber in prediabetes.

Diet is a central mechanism driving increased incidence of diabetes in young adults and current trends in rates of prediabetes [1,2,33]. Questions remain on the ideal diet targets to improve glycemic control and lower risk. Historically, a key focus has been on promoting a negative energy balance to facilitate weight loss, an approach that is efficacious compared to standard-of-care pharmaceutical intervention [62]. However, as in other chronic diseases, maintaining this negative energy balance is challenging. Therefore, there is interest in alternative approaches that are simple to conceptualize yet effective for improving glycemic control. From this perspective, decreasing total carbohydrates as in a ketogenic or carnivore diet is highly appealing. Carbohydrates are the nutrient responsible for increases in blood glucose; thus, their elimination would result in a decreased frequency of blood glucose spikes. Decreasing carbohydrates in this manner may also result in weight loss as individuals do succeed in losing weight on carbohydrate-restricted diets [63]. However, this may only mask underlying insulin resistance if fiber in the diet is key to improving glucose regulation. Indeed, recent evidence highlights the potential of fiber to facilitate microbiome modulation, resulting in improved glycemic control and response to pharmaceutical intervention [64]. However, this may only occur with foods high in intrinsic fiber, not isolated or synthetic fiber supplementation (eg, isolated psyllium husk) [65].

While fiber can be consumed in the context of low-carbohydrate diets, options that are both low in carbohydrate and high in intrinsic fiber are few. Furthermore, carbohydrate-containing fiber sources may be germane for reducing diabetes risk. In the empirical diabetes risk reduction diet, nuts (hazard ratio [HR] 0.57 for White adults and 0.65 for adults from minority groups) and cereal fiber (HR 0.9 for White adults and 0.95 for adults from minority groups) are both associated with reduced diabetes risk [12]. This is contrasted with red and processed meat, which has an HR of 1.34 for White adults and 1.14 for adults from minority groups [12]. Nuts are indeed low in carbohydrates, with 1.9 g of carbohydrates for every 1 g of fiber (ie, a 1.9:1 carbohydrate-to-fiber ratio) [66]. Some cereals, such as oats, have a higher carbohydrate content but still a relatively low carbohydrate-to-fiber ratio (6.7:1). Existing research shows that greater reliance on foods with less than a 10:1 ratio is associated with lower fasting glucose and HOMA-IR [66].

Cereal fiber can be consumed in isolated or synthetic forms present in specialty food products that may or may not have lower total carbohydrate content. Whether these have the same effects as intrinsic cereal fiber is not well known, and effects may be limited to cholesterol [67]. If the other food components present in intrinsic fiber sources (eg, polyphenols) are partially responsible for the protective glycemic effects, specialty food products may not prove advantageous. In their recent systematic review and meta-analysis, Colak et al [68] concluded that whole foods—as compared to supplemental sources—only improve HbA_1c versus other measures of glycemic control. However, most whole food–based studies included in their review appeared to be whole food sources of isolated or synthetic fibers (eg, “soy fiber in cookies”) vs food sources of intrinsic fiber (eg, soybeans). This is a vital distinction given the influx of products on the market that are markedly high in isolated or synthetic fiber.

Despite the potential benefits of eating more whole fiber, individuals may be reluctant to do so given that fiber is itself a carbohydrate and commonly coexists with digestible carbohydrates such as starch [30]. For example, per 1-cup serving, white beans contain 42.6 g of digestible carbohydrates and 13 g of fiber, whereas a food such as green beans has 4.8 g of digestible carbohydrates but only 3.5 g of fiber. Thus, if it is fiber intake—vs carbohydrates per se—that lowers risk of diabetes, a slightly higher carbohydrate intake to enable sufficient fiber may be necessary [69]. This is particularly true as fiber intake in excess of 30 g per day is associated with the greatest reduction in risk. In their systematic review and meta-analysis, Reynolds et al [69] observed a dose-response relationship between fiber intake and type 2 diabetes risk, including up to 45 g per day. Ingesting sufficient fiber via low–fiber density foods (such as green beans or nuts) may result in excessive calorie or total food intake, and fiber-dense sources (such as legumes) may be essential. The use of CGMs, as in this study, may help individuals observe the differences between acute increases in blood glucose and improved insulin sensitivity, key distinctions when adding fiber-containing foods to the diet. While the starch in foods such as beans may cause acute increases in glucose, this alone is not indicative of insulin resistance. Continuing to rely on foods such as these may improve long-term glucose regulation despite short-term increases in serum glucose levels [70].

Promoting greater intake of intrinsic fiber sources can also serve as an approach to behavior change. As outlined in the study by Ma et al [5] and others [8], by promoting greater intake of foods high in intrinsic fiber, participants are in essence merely working toward adoption of an overall healthy dietary pattern (eg, increasing intake of fruits, vegetables, whole grains, and beans or legumes) but via a singular focus on foods high in fiber. Existing data show that conventional advice to eat more plants, fruits, and vegetables or promote adoption of other disease-protective dietary patterns (eg, the Mediterranean diet) does not necessarily translate into improvements in fiber intake sufficient to meet levels that are associated with the highest disease risk reduction [71,72]. However, there is emerging evidence suggesting that greater fiber intake may be associated with greater improvements in multiple effects, such as sleep and mood, which may offer a greater return on investment, promoting and incentivizing long-term maintenance [73-75]. If intervention guidance is relatively simple and offers greater potential for perceived physiological change, it may promote intake in the long term.

While this study is innovative, it is not without limitations. First, as a single-arm trial, the effects of the intervention cannot be fully ascertained. Future work should include randomized trials, including comparators that feature either standard-of-care guidance (eg, DPP) or alternative diets that tap into the same psychosocial mechanisms via different dietary targets (eg, ketogenic or carnivore diet). These future trials should include more robust measures of glucose homeostasis, including oral glucose tolerance tests. Additionally, while participants are provided with food in this study, it is not a full outpatient feeding trial wherein all food is provided. This introduces variability; however, we are seeking to assess this variability in part via EMA, which could result in an intervention with greater ecological validity and potential for long-term effectiveness.

The results of this study will provide pivotal evidence on the utility of a fiber-focused intervention delivered via mHealth with biofeedback and home-delivered foods to lower risk of diabetes in young adults.