Production of nutrient‐enhanced millet‐based composite flour using skimmed milk powder and vegetables

Abstract The aim of this study was to develop a nutrient‐enhanced millet‐based composite flour incorporating skimmed milk powder and vegetables for children aged 6–59 months. Two processing methods were tested to optimize nutrient content and quality of millet‐based composite flour, namely germination for 0, 24 and 48 hr and roasting at 80, 100, and 140°C. The amount of ingredients in the formulation was determined using Nutri‐survey software. Germinating millet grains for 48 hr at room temperature significantly (p < 0.05) increased protein content (9.3%–10.6%), protein digestibility (22.3%–65.5%), and total sugars (2.2%–5.5%), while phytate content (3.9–3.7 mg/g) decreased significantly (p < 0.05). Roasting millet grains at 140°C significantly (p < 0.05) increased the protein digestibility (22.3%–60.1%) and reduced protein (9.3%–7.8%), phytate (3.9–3.6 mg/g), and total sugar content (2.2%–1.9%). Germinating millet grains at room temperature for 48 hr resulted in millet flour with the best nutritional quality and was adopted for the production of millet‐based composite flour. Addition of vegetables and skimmed milk powder to germinated millet flour significantly (p < 0.05) increased the macro‐ and micronutrient contents and the functional properties of millet‐based composite flour. The study demonstrated that the use of skimmed milk powder and vegetables greatly improves the protein quality and micronutrient profile of millet‐based complementary foods. The product has the potential to make a significant contribution to the improvement of nutrition of children in developing countries.

protein content of millet is inadequate for infant feeding and limiting in some essential amino acids like lysine, which is required for children's growth (FAO, 1995;Friedman, 1996). Although millet has considerably high iron content, its bioavailability is low (Sihag, Sharma, Goyal, Arora, & Singh, 2015). In addition, millet has low levels of βcarotene, vitamin C, zinc, calcium, sodium, and potassium (Queroz, 1991), which are required for child and infant feeding.
Thick porridge from millet flour is a major complementary food for infants and young children in most parts of Uganda (Temba, Njobeh, Adebo, Olugbile, & Kayitesi, 2016). The thick porridge from millet flour contains poor quality and inadequate protein and micronutrients required by weaning children. This is because it is nutritionally bulky and children have small volumes of the stomach. Therefore, they are not able to obtain adequate nutrients from the amount of porridge, and their stomachs are able to accommodate. Achieving a thin drinkable porridge requires the addition of copious amounts of water during preparation, which lowers the energy density (Adenike, 2012). However, processes such as sprouting/germination of millet grains activate endogenous enzymes that start modification of the millet grain constituents, causing changes in soluble sugars, protein, and fats that improve the nutritional quality, functional, and sensory properties of millet flour (Adenike, 2012). Consequently, starch (amylose and amylopectin) is degraded, resulting in a significant reduction in the viscosity and an increase in protein bioavailability during porridge preparation (Shaik, Carciofi, Martens, Hebelstrup, & Blennow, 2014). Germinating millet grains thus provides an easy way to improve the nutritional well-being of weaning babies.
Despite the improvement in nutritional and functional properties brought about by germination of millet grains, overall the millet-based diet remains inadequate in protein and micronutrients.
Addition of vegetables and skimmed milk powder is a promising option to improve both the macro-and micronutrient profile and functional properties of millet flour. This is because vegetables are rich in β-carotene, zinc, calcium, sodium, and potassium (Buttriss, 1989;Maass, 2014), while skimmed milk powder is rich in protein with balanced essential amino acids and high digestibility (Hoffman & Falvo, 2004). The aim of this study was therefore to develop a nutrientenhanced millet-based composite flour incorporating skimmed milk powder and vegetables.

| Selection of raw materials
The materials used in this study included millet grains, pumpkin seeds, carrots, cowpea leaves, and skimmed milk powder. These raw materials were selected because of their superiority in the macro-and micronutrients essential for children aged 6-59 months.
Skimmed milk powder has a high protein content with balanced essential amino acids and is stable during storage due to a low fat content (Hoffman & Falvo, 2004). It also increases calcium and iron bioavailability due to its high protein content (Gibson, Bailey, Gibbs, & Ferguson, 2010). Millet has high iron and calcium contents required for infant blood transportation and bone formation, respectively. In addition, it is a major source of energy for infant growth. Although iron in millet is less available, the addition of cowpea leaves and carrots and processing methods was suggested ways of improving the iron bioavailability (Christides, Amagloh, & Coad, 2015). Pumpkin seeds are rich in zinc with 7.5%-14% bioavailability of the zinc requirements (Glew et al., 2006), which is essential for child development, while carrots are rich in β-carotene (Singh, Kawatra, & Sehgal, 2001), which is essential for immune development in infants aged 6-59 months. Cowpea leaves are rich in protein, iron, and calcium and have low levels of antinutrients compared with other proteinrich plant sources (Chikwendu, Igbatim, & Obizoba, 2014).

| Source of raw materials and laboratory reagents
Millet, pumpkin seeds, carrots, and cowpea leaves were purchased from Kisenyi-Market, Kampala, Uganda, while skimmed milk powder was purchased from Pearl Dairies, Mbarara District, Uganda. All the materials were delivered to the laboratory at the School of Food Technology, Nutrition and Bioengineering, Makerere University for further processing. Laboratory reagents were purchased from Neo Faraday Laboratory Supply, Kampala, Uganda.

| Preprocessing of millet grains
Millet grains were divided into two equal portions. One portion was germinated and the other roasted. Germination time was selected based on the sprouting/germination time of millet grains (1-2 days) (Saleh, Zhang, Chen, & Shen, 2013), while roasting temperatures were selected based on methods of Adebiyi, Adeyemi, and Olorunda (2002).

| Preparation of vegetables
Cowpea leaves and carrots were washed with running tap water to remove surface soil. The cowpea leaves and carrots were blanched in a water bath (Grants Instrument Ltd, Shepreth, UK) maintained at 80°C for 10 min and 95°C for 5 min, respectively. After blanching, carrots were shredded using a hand grater with holes of a diameter of 0.6 cm.
The shredded carrots and cowpea leaves were dried for 6 hr in an electric cabinet dryer (B.MASTER, Italy) set at 55°C. Dried carrots and cowpea leaves were then milled into fine powders using a locally fabricated hammer mill. Pumpkin seeds were cleaned and roasted at 140°C for 20 min using an infrared Food oven (GU-6, New Zealand). Dry pumpkin seeds were blended into a fine powder using an electric blender (Lilaram Manomal and Sons, India).

| Millet grains processing conditions
Germination time and roasting temperatures were varied as shown in Table 1. Regression analysis was used to determine the influence of germination time and roasting temperature on the nutritional quality of the millet flour. The protein digestibility, protein content, total phytates, and total sugars were expressed individually as functions of the germination time and roasting temperature. The data were fitted to the approximating linear model using Equations (1) and (2). where Y = Response function, β 0 = Constant (the intercept), β 1 or β 2 = Coefficient of the linear effects, X 1 and X 2 = Linear effects of germination time and roasting temperature, respectively.

| Formulation of the composite flour
Nutri-survey software, 2007 was used to estimate the optimum amounts of skimmed milk and vegetable powders to add to millet flour in order to meet the protein (13 g/day) and energy (1046 and 902 kcal/day for boys and girls, respectively) requirements for children of age 6-59 months (WHO, 2012). The details of the various combinations used to reach the optimum levels of millet flour, skimmed milk, and vegetable powders are indicated in Table 2.

| Proximate composition and energy estimation
The millet and millet-based composite flours were analyzed for moisture, crude protein, crude fat, crude fiber, and ash contents according to the method described by AOAC (2010). Carbohydrate was determined by difference, and energy content was determined using the Atwater factor (carbohydrate and protein values were each multiplied by 4 kcal/g, whereas fat values were each multiplied by 9 kcal/g).

| Determination of minerals
The amount of calcium, zinc, iron, magnesium, and copper in the composite flour was measured by an atomic absorption spectrophotometer according to the method of Hernández-Urbiola, Pérez-Torrero, and Rodríguez-García (2011). Calibration equations were derived, and concentrations of calcium, zinc, iron, sodium, and potassium were expressed as mg/100 g.
where y = intercept on y-axis; D = Dilution factor; 10 = a conversion factor from mg/kg to mg/100 g; V = volume of sample in ml.

| Determination of β-carotene
The β-carotene content of flour was determined according to Rodriguez-Amaya and Kimura (2004). Carotenoid and the β-carotene contents were expressed as μg/g and μg RAE, respectively.

| Determination of total phytates
A modified method of Ijarotimi and Babatunde (2013) was used for the determination of phytate content in millet flour. (1)

| Determination of protein digestibility
Protein digestibility was determined by the porcine pepsin method as adapted by Gomez, Obilana, Martin, Madzvamuse, and Monyo (1997).

| Determination of total sugars
Total sugars were determined according to the method described by Nielsen (2010).

| Determination of physical properties of millet-based composite flour
The bulk density, water, and oil absorption capacities of the millet flours were determined according to the method described by Onwuka (2005).

| Determination of swelling capacity and solubility of millet-based composite flour
The method described by Leach et al. (1959) was used to determine swelling power and solubility of the sample. The swelling power and solubility of the samples were expressed as %.

| Pasting properties
Pasting characteristics of the porridge from the composite flour were determined with a Rapid Visco Analyzer (Perten Instruments AB, Kungens Kurva, Sweden) according to Ikegwu, Okechukwu, and Ekumankana (2010). Peak viscosity, trough, breakdown, final viscosity, setback, peak time, and pasting temperatures were read from the pasting profile with the aid of thermocline for windows software connected to a computer (Newport Scientific, 2001). The viscosity was expressed in centipoises (cP).

| Contribution of porridge from millet-based composite flour to RDA
Percentage contribution to recommended dietary allowance was expressed as % of RDA.
where X is the amount of nutrient analyzed and Y is the RDA for a given nutrient/variable.

| Statistical analysis
All determinations were performed in duplicates and subjected to statistical two sample t test analysis of variance (ANOVA) using XLSTAT software version 2017 to determine variation between means. Significance variation was accepted at p < 0.05. Simple linear regression analysis was used to determine the influence of germination time and roasting temperature on the nutritional quality of the millet flour.

| Effect of roasting and germination on the nutrient content and quality of millet flour
The effect of roasting and germination on the nutrient composition of millet flour as predicted by the linear regression models is indicated in Table 3. Germination time had a positive linear effect on the protein content, total sugar content, and protein digestibility as well as a negative linear effect on the total phytates in millet flour. According to the results in Table 3, an increase in germination time significantly (p < 0.05) increased the protein content, protein digestibility, and total sugar content of millet flour by 0.03%, 0.79%, and 0.07%, respectively, and reduced the total phytates by 0.005 mg/g. Roasting temperatures had a positive linear effect on the protein digestibility and a negative linear effect on the total phytates, protein, and total sugar contents of millet flour. An increase in roasting temperature significantly (p < 0.05) reduced the total phytates by 0.003 mg/g, protein content, and total sugar content of millet flour each by 0.01%. However, an increase in roasting temperature increased the protein digestibility of millet flour by 0.34%.
Swelling power (%) = weight of sediment paste (g) × 100 weight of sample (g) × (100 − %Solubility) ,   Figure 1 shows that germinating millet grains for 48 hr significantly (p < 0.05) increased the protein content (9.3%-10.6%), total sugars (2.2%-5.5%), and protein digestibility (22.3%-65.5%). Results in Figure 2 show that roasting millet at 140°C significantly (p < 0.05) reduced total sugars (2.2%-1.9%), protein content (9.3%-7.8%), and phytates (3.9-3.6 mg/g) but increased protein digestibility (22.3%-60.1%). The study findings indicate that germination for 48 hr resulted in millet flour with a high protein content (10.6%), protein digestibility (65.5%), total sugars (5.5%), and low levels of phytates (3.7 mg/g) and was therefore taken as the optimal procedure to process the flour that was used in the formulation of the composite flour. The increase in the total sugars during germination of millet grains is due to the hydrolysis of starch into shorter chain sugars by amylases. The results of this study agree with those reported by other researchers. Coulibaly and Chen (2011) reported a drastic increase in total soluble sugars (1%-13%) in foxtail millet germinated for 6 days.

| Optimal processing conditions of millet flour
The significant increase in protein content observed in flour from germinated millet grains is attributed to an increased water activity that activates hydrolytic enzymes (Nonogaki, Bassel, & Bewley, 2010) as well as compositional changes associated with dry matter loss following the degradation of other constituents (Abdelhaleem, El Tinay, Mustafa, & Babiker, 2008). The physical mechanisms and biochemical reactions that explain the observed increase in bioavailability of protein content during the germination process are strongly associated with the morphology of the finger millet seed (Hejazi & Orsat, 2016). In the finger millet, endosperm F I G U R E 1 Effect of germination time on the protein content, protein digestibility, total sugars, and total phytates in millet flour F I G U R E 2 Effect of roasting temperature on the protein content (a), protein digestibility (b), total sugars (c), and total phytates (d) in millet flour represents the largest portion of the grain, which consists of an aleurone layer and three distinct starchy sections: peripheral, corneous, and floury endosperms. The peripheral endosperm contains small and tightly packed cells of protein bodies that are embedded in fiber-starch-protein matrices (Belton & Taylor, 2002). The observed increase in the crude protein level during germination was the outcome of the activation of the plant α-amylase (Traoré, Mouquet, Icard-Vernière, Traore, & Trèche, 2004). The activity of α-amylase resulted in the starch granules breakdown, which led to the release of the protein from the packed cells and consequently, increased its bioavailability. Findings from this study are in agreement with those of Chaturvedi (2015), who also reported a drastic increase in protein content from 12.5%-58.6% in foxtail millet germinated for 48 hr.
The increase in protein digestibility observed in flour from germinated millet grains is due to phytic acid hydrolysis, which is known to interact with proteins to form complexes as well as degradation of long protein chains by proteases whose activity increases during germination of cereals (Hassan et al., 2006;Mbithi-Mwikya et al. 2000). Osman (2007) observed a similar trend in germinated Lablab beans.
Reduction in the total sugar levels and protein content during roasting of millet grains could be attributed to Maillard reactions and caramelization, which occur at higher temperatures (Baker et al., 2003). During roasting, sugars, and amino acids, in particular lysine because of its free ∈-amino groups, undergo Maillard reactions, thus reducing protein and sugar contents (Gilani, Xiao, & Cockell, 2012) of millet flour. In addition, roasting reduces the protein content of millet flour due to protein denaturation (Kavitha & Parimalavalli, 2014). Table 4 shows the proximate composition of millet-based composite flour on a dry weight basis. Results in Table 4 show that addition of vegetables and skimmed milk powders to the millet, significantly (p < 0.05) increased the crude protein, crude fat, crude fiber, ash, and gross energy of millet-based composite flour by 7.3%, 2.1%, 2.5%, and 1.6%, respectively. These results further indicated that moisture content and carbohydrate content of the millet-based composite flour significantly (p < 0.05) decreased by 1.1% and 9.9%, respectively.

| Proximate composition of millet-based composite flour
The In case, the product is to be fed to malnourished children, and further modification will be required to reduce the fiber content to the required standard. The significant (p < 0.05) increase in the ash con-  (Table 4). The results of the energy content from this study are in agreement with the findings of Balasubramanyam and Lokesh (2000), who reported that supplementary foods prepared from cereals and pulses provide 10%-30% proteins and 350-380 kcal energy. (2.1-4.2 mg/100 g), copper (0.5-0.9 mg/100 g), and calcium (143.6-667.8 mg/100 g) contents was observed in the millet-based composite flour. The results in Table 5 further indicate that the increases in iron (3.4-3.6 mg/100 g) and magnesium (4.3-4.4 mg/100 g)

| Mineral and vitamin A (RAE) content of milletbased composite flour
were not significant (p > 0.05). The findings also indicate that the vitamin A content of millet-based composite flour was significantly (p < 0.05) higher (641 μg RAE/100 g) than that of millet flour (15.5 μg Based on compositional analysis (Table 5), daily consumption of 100 ml porridge from the millet-based composite flour will meet the RDA of children 6-59 months.  Table 5. Zinc, copper, and vitamin A are nontoxic in the body (Food and Nutrition Board, 1989), and therefore, their high levels in the millet-based composite flour have no health concern.

| Contribution of mineral and vitamin A content of porridge prepared from the millet-based composite flour toward RDA for children aged 6-59 months
Therefore, adoption of the millet-based flour and its proper preparation may greatly contribute to the reduction in mineral deficiencies among children aged 6-59 months.
3.6 | Contribution of energy, protein, and fat content of porridge prepared from 100 g of milletbased composite flour in 700 ml of water toward RDA for children aged 6-59 months Table 7 shows the contribution of millet-based composite flour to the RDAs of energy, protein, and fat for children aged 6-59 months. The findings show that the protein contribution to the RDA reduced with an increase in the age of children. For children 0.5-1 year, the porridge from millet-based composite flour would meet 109% of the RDA for protein, but it meets only 63.8% for children aged 4-6 years. Energy and fat contribution followed the same trend as that of protein, but all values were below 100% (Table 7).
The high contribution of the millet-based composite porridge to RDAs for protein and energy is due to high concentrations of these TA B L E 5 Mineral (mg/100 g) and vitamin A (μg RAE/100 g) content of millet flour and millet-based composite flour TA B L E 6 Contribution (%) of mineral and vitamin A content of porridge prepared from 100 g of millet-based composite flour in 700 ml of water toward RDA for children aged 6-59 months a nutrients in the composite flour (Table 5). In addition, the contributions of protein that are above the RDA are nontoxic to the body (Food and Nutrition Board, 1989; Nutrition Board and Institute of Medicine, N. A., 2011) because it was slightly above the protein requirement. However, it is recommended that protein intake should not be more than twice the RDA for protein (Food and Nutrition Board, 1989). Reduction in the contribution of energy, protein, and fat to the RDA with an increase in age is due to an increase in the body needs during growth. For example, energy is needed for metabolic activities and body maintenance, while protein is for growth and development in children. Fats are essential in the body for; calorie supply, brain development, absorption and transportation of vitamins A, D, E and K. In order to meet the protein, fat, and energy RDAs of the older children, intake of more than 100 ml of the milletbased composite porridge is recommended.  This observation is consistent with the reports of Chandra, Singh, and Kumari (2015), who observed an increase in OAC of composite flours prepared by blending wheat flour with rice flour, mung bean flour, and potato flour from 146% to 156%. The values observed in this study were higher than those of sweet potatoes flour (10%-12%) (Choi and Yoo, 2008) but lower than those in lupin seed flour (167%) (Sathe et al., 1982).

| The physical properties of millet-based composite flour
The high WAC of millet composite flour is due to the high protein content resulting from the addition of cowpea leaf and skimmed milk powder as well as to the changes in the quality and quantity of proteins in the flour upon germination of millet grains (Sreerama, Sashikala, Pratape, & Singh, 2012 TA B L E 8 Physical properties of millet flour and millet-based composite flour soluble in water while preparing porridge. The values of WAC obtained in this study were higher than the values reported for taro flours (red-180%; white-166%; and nive-150%) (Tagodoe and Nip, 1994), soybean flour (130%) (Lin et al., 1974), fluted pumpkin seed flour (85%) (Fagbemi and Oshodi, 1991), and sweet potato flours (red-24% and white-26%) (Osundahunsi et al., 2003), probably because of the high protein content of the millet-based composite flour.
However, the values obtained were in the range of those reported by Fasasi, Eleyinmi, and Oyarekua (2007), which were 226%-270% in processed millet flour samples.
Incorporation of vegetable and skimmed milk powders did not change the bulk density of millet composite flour (Table 8). This could be attributed to the low weight of the added powders in comparison with millet flour. In contrast, Edema, Sanni, and Sanni (2005) reported a decrease in bulk density of maize with increasing soy supplementation. This is probably because of the high weight of soy flour. The low values of bulk densities of millet composite flour make it suitable for high nutrient density formulations of foods. The setback or viscosity of the cooked paste is the viscosity after cooling to 50°C. The extent of increase in viscosity on cooling to 50°C reflects the retrogradation tendency (Chibuzo, 2012), a phenomenon that causes the paste to become firmer and increasingly resistant to enzyme attack (Horstmann, Lynch, & Arendt, 2017). It thus has an effect on digestibility. Higher setback values are synonymous to reduced paste digestibility (Shittu & Adedokun, 2010), while lower setback during cooling of the paste indicates a lower tendency for retrogradation and subsequently higher digestibility (Sandhu, Singh, & Malhi, 2006). The low setback value (16.67 cP) for the millet-based composite flour indicates that its paste would have a higher stability against retrogradation (Mazurs, Schoch, & Kite, 1957) than millet flour whose setback value was 77 cP. It also implies that the porridge when consumed by children will be easy to digest.

| Effect of temperature on the solubility of millet-based composite flour
The pasting temperature of millet-based composite flour (55.7°C) was significantly (p < 0.05) lower than that of millet flour (67.9°C).
The pasting temperature provides an indication of the minimum temperature required for cooking the flours (Shimelis, Meaza, & Rakshit, 2006). The low pasting temperature of millet-based composite flour implies that less energy is required for cooking porridge from milletbased composite flour than for flour from nongerminated millet grains. F I G U R E 5 Rapid Visco-Analyzer pasting curves for millet flour and milletbased composite flour

ACK N OWLED G EM ENTS
We are grateful to Netherlands Organization for Scientific Research/WOTRO Science for Development who provided financial assistance that enabled this work to be done. We also acknowledge the technical assistance offered by the laboratory team at the School of Food Technology, Nutrition and Bioengineering, Makerere University.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.