Sweet potato is a key food and nutrition security crop in Malawi and across sub-Saharan Africa, especially biofortified orange-fleshed varieties that address vitamin A deficiency. However, productivity and market performance remain low due to limited access to quality planting materials, weak postharvest systems, and fragmented value chain governance. Emerging digital innovations such as machine learning (ML), hyperspectral imaging (HSI), and remote sensing present new opportunities to improve productivity, efficiency, and inclusivity across the sweet potato value chain.
Objective
This systematic review synthesizes evidence on how data-driven and gender-responsive innovations can enhance productivity, equity, and sustainability within the sweet potato value chain in Malawi and comparable African contexts.
Methods
Peer-reviewed studies were systematically reviewed following PRISMA guidelines. Twenty-six articles published between 2015 and 2025 were selected from Google Scholar and ScienceDirect. Data extraction captured study characteristics, focus areas, and outcomes. Study quality and bias were assessed using the Newcastle–Ottawa Quality Assessment Scale.
Results
Of the reviewed studies, 27% focused on production, utilization, and governance, revealing gaps in coordination, policy alignment, and value addition. About 43% examined ML and HSI applications for yield prediction, quality grading, and spatial mapping, reporting 80–95% accuracy in predicting physical and nutritional attributes. Only 17% explicitly addressed gender and inclusion, showing persistent inequities in digital participation. A smaller subset (13%) proposed frameworks combining AI, participatory governance, and policy tools for inclusive food systems. Collectively, the evidence highlights rapid advancement in digital agriculture but limited integration with gender, governance, and capacity-building dimensions.
Conclusion
Data-driven tools hold transformative potential to enhance productivity, quality, and market systems in sweet potato value chains. To realize this potential, digital agriculture must be embedded in inclusive, gender-responsive, and context-specific frameworks that strengthen research–policy–practice linkages, data infrastructure, and institutional collaboration for resilient and equitable food systems.
data sciencedata-driven agriculturedigital transformationMalawisweet potatoThe author(s) declared that financial support was received for this work and/or its publication. The authors declare that the African German Center for Sustainable and Resilient Food Systems and Applied Agricultural and Food Data Science (UKUDLA) provided partial funds to publish this systematic review article.section-at-acceptanceLand, Livelihoods and Food SecurityIntroduction
Sweet potato (Ipomoea batatas) is a major root and tuber crop in sub-Saharan Africa, valued for its adaptability to marginal environments, short growing cycle, and contribution to household food security (Cartabiano-Leite et al., 2020). Once regarded primarily as a subsistence crop, sweet potato has increasingly assumed a strategic role in smallholder farming systems as a source of both food and income. Particularly, under conditions of climate variability and declining reliability of maize-based systems hence contributing to Sustainable Development Goals 1(No Poverty), 2 (Zero Hunger) and 13 (climate action) (Afzal et al., 2021). Across the region, the crop supports millions of rural households by providing calories during lean seasons and by offering opportunities for local market participation and value addition (Islam, 2024).
In Malawi, sweet potato is among the country’s most widely cultivated root crops, alongside cassava and Irish potato, and is consumed across socio-economic groups. Its ability to perform well under low-input conditions and variable rainfall has made it a critical component of household food security strategies, especially in drought- and flood-prone districts. Beyond subsistence use, sweet potato functions as an important seasonal cash crop, with smallholder farmers selling surplus harvests in local markets to meet immediate household needs such as food purchases, healthcare, and school fees (Motsa et al., 2015). As a result, the crop occupies a dual role in Malawi’s food system, simultaneously supporting subsistence consumption and livelihood generation.
Nutritionally, orange-fleshed sweet potato (OFSP) has been widely promoted in Malawi and across the region to address micronutrient deficiencies, especially vitamin A deficiency among women of the reproductive age and children (Kapalasa et al., 2022). OFSP varieties are rich in beta-carotene and have been incorporated into national nutrition and agricultural programs, including school feeding and community-based nutrition interventions. Evidence shows that integration of OFSP into farming and consumption systems can improve dietary diversity and contribute to better nutrition outcomes, reinforcing the crop’s relevance beyond caloric provision (Abidin et al., 2015).
Despite its agronomic and nutritional advantages, sweet potato production in Malawi is characterized by recurrent seasonal oversupply. Harvests are typically concentrated once or twice per year, leading to flooded markets, sharp price declines, and substantial post-harvest losses. Limited storage infrastructure, inadequate processing capacity, and weak market coordination exacerbate these challenges, with post-harvest losses in some rural markets estimated to reach 30–40%. These losses translate into reduced incomes for farmers and traders and undermine the crop’s potential to contribute meaningfully to livelihood improvement and national food and nutrition security.
Gender dynamics further shape outcomes along the sweet potato value chain. Women and youth constitute a significant proportion of producers and traders. However, they are often concentrated in lower-value chain nodes, such as production and informal retailing, while men dominate more profitable segments including commercial seed systems, aggregation, and processing (Munyaka et al., 2025). Limited access to assets, decision-making power, and market information constrains women’s and youth’s ability to benefit from commercialization opportunities, reinforcing the treatment of sweet potato as a subsistence crop rather than a viable enterprise.
Although various interventions have sought to promote value addition, market integration, and inclusive business models for sweet potato in Africa, evidence remains fragmented regarding how oversupply can be systematically addressed to improve livelihoods, particularly in Malawi (Yadav and Sidana, 2023). Moreover, while data-driven and digital approaches are increasingly applied across agricultural value chains, their role in informing production planning, post-harvest management, and market coordination in sweet potato systems has not been comprehensively analyzed. This systematic review addresses this knowledge gap by examining existing evidence on sweet potato oversupply and livelihood outcomes in Malawi, using data-driven and gender-responsive approaches as an analytical lens. Specifically, the review synthesizes evidence on production patterns, post-harvest losses, value chain challenges, and livelihood impacts. The review further assesses how analytical and digital tools have been applied to support more resilient and inclusive sweet potato value chains.
Conceptual framework
This systematic review was guided by an integrated conceptual framework combining the Value Chain Analysis and Sustainable Livelihoods perspectives to examine sweet potato production, processing, and consumption in Malawi. The Value Chain framework, drawing from Porter’s Value Chain and Gereffi’s Global Value Chain approaches, provides a systematic lens for understanding the sequential activities among actors in the sweet potato value chain (Zamora, 2016; Gereffi and Fernancez-Stark, 2011). This perspective enabled the identification of bottlenecks and areas where coordination, data-science tools and policy interventions could enhance efficiency and equity through data-driven decision-making. It also enabled understanding of how gender dynamics influence the uptake and effectiveness of these tools, particularly among women and youth who dominate informal processing and marketing activities.
Figure 1 illustrates the conceptual framework for the sweet potato value chain in Malawi, linking empirical, data science-informed observations to strategies for strengthening key players and nodes of the value chain.
Data science integrated conceptual framework for sweet potato value chain in Malawi.
Flowchart depicting the agricultural value chain. It starts with "Inputs/Resources" including seeds, labor, tools, and knowledge, leading to "Production and Aggregation" by farmers and traders. "Processing and Value Addition" follows with flour and chips, then "Markets/Consumption" targeting local and urban buyers. Additional arrows indicate factors like "Household and farm assets," "Information and ICT Structures," and "Gender and Youth Inclusion."MethodsStudy design and reporting framework
This study employed a systematic literature review to analyze evidence on sweet potato oversupply and livelihood outcomes in Malawi, using data-driven and gender-responsive approaches as analytical lens. The review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines to ensure transparency, reproducibility, and methodological rigor (Page et al., 2021). A systematic review design was selected to enable structured identification, screening, and synthesis of empirical studies examining production patterns, post-harvest losses, value chain dynamics, applied agricultural data science and livelihood implications associated with sweet potato systems.
Literature search strategy
A comprehensive literature search was conducted across peer-reviewed and selected grey literature sources. Electronic databases searched included Google Scholar and ScienceDirect, which were used to identify open-access studies relevant to agriculture, food systems, and digital innovations. The search covered literature published between 2015 and 2025 to capture recent developments in digital and analytical approaches relevant to agricultural value chains, while allowing inclusion of seminal studies published earlier where they provided foundational context. The literature search was conducted using standardized Boolean operators to ensure consistency and reproducibility across databases. Core search terms related to data science, sweet potato, value chains, and livelihoods were combined using “AND,” while synonymous or closely related terms were linked using “OR” for example, “data science” OR “machine learning” OR “digital agriculture.” Truncation and phrase searching were applied where supported to capture variations in terminology.
In addition to database searches, backward and forward citation tracking was employed to enhance coverage. Reference lists of all included articles were manually screened following backward citation tracking to identify additional relevant studies not captured in the initial search. Forward citation tracking was also conducted using Google Scholar to identify more recent publications that cited the included studies. This iterative approach helped mitigate the risk of omitting relevant literature, particularly given the emerging and interdisciplinary nature of data science applications in agricultural value chains.
Eligibility criteria
Predefined inclusion and exclusion criteria were applied to ensure relevance and consistency. Studies were included if they focused on sweet potato production, post-harvest handling, value chains, or market dynamics; examined livelihood, income, gender, or food security outcomes; and applied analytical, digital, or data-driven approaches to inform decision-making, coordination, or prediction within agricultural systems.
Primary empirical studies conducted in Malawi and in comparable Sub-Saharan African contexts were included to allow contextual transferability. Studies published in English were considered.
Studies were excluded if they did not address sweet potato systems; lacked empirical evidence; or consisted solely of review articles, perspective papers, or opinion pieces. Review articles were used only for background and conceptual framing and were not included in the analytical dataset.
Study selection process
All identified studies were imported into a screening database, and duplicates were removed prior to screening. Titles and abstracts were first screened against the eligibility criteria, followed by full-text screening of potentially relevant articles. The number of records identified, screened, excluded, and included is presented in the PRISMA flow diagram (Figure 2).
PRISMA flow diagram for the systematic review.
Flowchart titled "Identification of studies via databases" showing the stages of study selection. Identification stage: 17,400 records identified, 14,765 removed, 300 screened. Screening stage: 179 excluded, 121 sought for retrieval, 57 not retrieved. Eligibility stage: 64 assessed, 38 excluded. Inclusion stage: 26 studies included.Data extraction and synthesis
Data from included studies were extracted using a standardized extraction template aligned with the review objectives and conceptual framework. Extracted variables included publication year, country, study design, value chain focus, analytical or digital tools applied, gender considerations, and reported livelihood outcomes.
A thematic synthesis approach was used to analyze the data. Quantitative summaries were generated to describe trends by publication year, geographic focus, and methodological approach. Qualitative findings were coded inductively to identify recurring themes related to oversupply drivers, post-harvest losses, value chain coordination, gender dynamics, data science and livelihood impacts. To enhance reliability, data extraction and coding were independently cross-checked by all the authors (MC, PF, SN, JM, PN, JK, JN and KM) and discrepancies were resolved through discussion and consensus. Findings were synthesized narratively and supported by tables to illustrate relationships between analytical approaches, value chain challenges, and livelihood outcomes in sweet potato systems.
ResultsSelected studies
The literature search identified records across electronic databases and supplementary sources. Following duplicate removal, titles and abstracts were screened for relevance, and potentially eligible articles underwent full-text review. After applying the predefined inclusion and exclusion criteria, 26 studies were included in the final synthesis. Reasons for exclusion at the full-text stage were presented in the exclusion and inclusion criteria section. The study selection process is summarized in the PRISMA flow diagram (Figure 2).
Characteristics of included studies
The 26 included studies were published between 2020 and 2025 and covered diverse geographic contexts, including Malawi, other Sub-Saharan African countries, and selected studies from Asia, Europe, and North America with methodological relevance to sweet potato systems. Study designs included quantitative modelling, laboratory-based trials, remote sensing analyses, mixed-methods studies, and applied machine learning experiments. The analytical focus ranged from yield prediction and quality assessment to supply-chain optimization and post-harvest loss reduction.
Synthesized thematic findingsSweet potato production, utilization, and value chain governance
Seven studies (27%; n = 26) examined sweet potato production systems, utilization patterns, and value chain governance. These studies documented smallholder-dominated production systems, limited mechanization, and reliance on rainfed agriculture. Constraints frequently reported included limited access to improved planting material, weak post-harvest handling practices, and inadequate value addition infrastructure (Kawaye et al., 2020; Munyaka et al., 2025; Gholami et al., 2022). Several studies also identified coordination gaps among research institutions, extension services, and market factors affecting value chain efficiency.
Data-driven tools for real-time decision-making
Eleven studies (43%; n = 26) focused on the application of data-driven tools such as machine learning, hyperspectral imaging, and remote sensing (Coussement et al., 2024; Ahmed et al., 2025). These studies applied predictive models for yield estimation, quality grading, land-use mapping, and early detection of stress factors. Approaches included explainable artificial intelligence, UAV- and satellite-based modelling, and integration of climatic and agronomic datasets. Several studies reported the use of dashboards and early-warning systems to support decision-making across production, processing, and marketing stages (Aworka et al., 2022; Carbajal-Carrasco et al., 2024; Coussement et al., 2024; Kurek et al., 2023).
Gender-inclusive, data-informed interventions
Only four studies (17%; n = 26) explicitly addressed gender within data-driven agricultural interventions. These studies showed the importance of designing digital tools that consider differential access to technology, literacy levels, and participation of women and youth (Van Vugt and Franke, 2022; Mungai et al., 2022). Evidence from these studies emphasized participatory design and inclusive data governance as mechanisms to improve equitable access to benefits derived from digital agriculture innovations (Gholami et al., 2022; Gereffi and Fernancez-Stark, 2011).
Co-designing data-driven theoretical frameworks for inclusive food systems
Three studies (13%; n = 26) focused on conceptual or theoretical frameworks integrating data-driven tools within sustainable and inclusive food systems. These studies proposed combining machine learning, simulation models, and policy instruments to optimize supply chains, reduce post-harvest losses, and support adaptive planning (Munyaka et al., 2025). Frameworks emphasized participatory governance and prioritization of underutilized crops such as sweet potato within digital agriculture strategies (Diop et al., 2023; Nowakowski et al., 2024).
Risk of bias and study quality assessment
All included studies were assessed for methodological quality using an adapted Newcastle–Ottawa Quality Assessment Scale for cross-sectional studies. Of the 26 studies assessed, the majority were rated as high quality (scores 7–9), while the remainder were rated as moderate quality (scores 4–6). No studies were classified as low quality. Variability in study design, sample size, and validation procedures was observed across studies. Detailed quality assessment scores and justifications are presented in Table 1 which was submitted as a separate file (Alamu et al., 2020; Kpienbaareh et al., 2024; Dubey et al., 2024; Kawaye and Hutchinson, 2020).
Newcastle–Ottawa quality assessment scale for cross-section studies.
Author(s)
Year
Country of study
Study focus
Selection (0–4)
Comparability (0–2)
Outcome (0–3)
Total (0–9)
Quality rating
Comments
Coussment et al.
2024
Bangladesh
Explainable AI for enhanced decision-making.
4
2
3
9
High
Strong design, clear validation data, reproducible AI pipeline.
Lui et al.
2024
China
ML prediction of sweet potato traits (environmental factors)
4
2
3
9
High
Comprehensive dataset and controls for environmental confounders.
Nowakowski et al.
2021
Poland
Crop type mapping with transfer learning
3
1
3
7
High
Good spatial validation; limited comparability variables.
This section discusses the findings in themes that came up during the review process in relation to the topic at hand. These themes include; sweet potato production, oversupply and value chain constraints, contribution of data-driven tools to decision-making and value chain optimisation and gender inclusion gaps in data-driven sweet potato systems. Following the systematic review of systematic review of 26 studies to examine how data-driven approaches can address sweet potato oversupply and improve livelihoods. Special attention was given to governance, gender inclusion, and value chain integration. The findings show irregularities in the application of data science across the sweet potato value chain. The findings also indicate that important value chain nodes such as production and quality analytics receive more attention while other nodes such as post-harvest coordination, gender equity and systems-wide integration receive limited attention.
Sweet potato production, oversupply, and value chain constraints
The systematic review found that sweet potato production in Malawi and Sub-Saharan Africa is dominated by smallholder farmers operating under rainfed systems with limited mechanization and post-harvest infrastructure (Low et al., 2017). Agro-ecologically, sweet potato thrives across Malawi from lakeshore plains to highlands due to its adaptability ensuring food security and income diversification (Longwe et al., 2023; Gelaye, 2024). Studies addressing production and governance (27% of reviewed articles) consistently highlighted yield variability, weak market coordination, and inadequate value addition as key contributors to seasonal oversupply and post-harvest losses. These findings align with broader evidence that perishable root crops are particularly vulnerable to inefficiencies at aggregation, storage, and processing nodes. These inefficiencies at such key nodes often lead to failure to translate surplus production into income gains (Munyaka et al., 2025). The reviewed articles also show that oversupply is not solely a production issue but a systemic value chain challenge. Limited coordination among producers, traders, processors, and extension services constrains timely market responses, reinforcing cycles of excessively abundant supply and waste. These structural constraints show the need for integrated, data-informed planning rather than isolated productivity interventions.
Contribution of data-driven tools to decision-making and value chain optimisation
A substantial proportion of the reviewed literature (43%) focused on data-driven tools such as machine learning, remote sensing, and hyperspectral imaging (Liu et al., 2024; Ramírez et al., 2023). Together, these studies show strong potential for improving yield estimation, quality assessment, and spatial targeting of production. Predictive models integrating climatic, soil, and agronomic data were shown to enhance decision-making accuracy at farm and research levels. However, the review also reveals a concentration of data science applications at upstream value chain nodes. These nodes include production and phenotyping, with comparatively limited attention to downstream challenges such as post-harvest handling, market forecasting, and logistics coordination (Wang and Hsu, 2024). Only a small subset of studies (14%) addressed supply chain optimisation and loss reduction through integrated modelling. This imbalance shows that while data science tools are technically mature, their application remains fragmented across the sweet potato value chain in Malawi and other countries in the Sub-Saharan region.
The systematic review shows that if efforts are directed towards increasing digital literacy, platform accessibility, expanding predictive analytics to include demand forecasting, storage optimisation, and processing capacity planning, there can be greater chances to translate productivity gains into livelihood improvements for all the key players in the sweet potato value chain.
Gender inclusion gaps in data-driven sweet potato systems
Across the reviewed literature, gender and social inclusion emerged as a consistently underrepresented dimension of data-driven interventions in sweet potato value chains. Only 17% of the included studies explicitly incorporated gender considerations into either the design or evaluation of digital tools. Where gender was addressed, the evidence suggested that most data-driven applications implicitly assumed male-dominated access to technology, literacy, and decision-making authority (Kurek et al., 2023). As a result, digital tools for yield prediction, quality assessment, and market intelligence were frequently designed without accounting for gender-differentiated access to mobile devices, digital skills, and control over production and income (Su and Xue, 2021; Shoffe and Johnson, 2024).
The limited use of gender-disaggregated data across the reviewed studies further constrained the ability of data-driven models to capture differentiated livelihood outcomes. Few (17%) studies disaggregated users, adopters and beneficiaries by gender or age, and none systematically examined how algorithmic outputs or decision-support tools affected women and youth differently from men. This gap is particularly significant given that women and youth are widely documented within the reviewed literature as dominant actors in informal production, processing, and marketing segments of sweet potato value chains. The absence of their perspectives in data collection and tool design therefore risks reinforcing existing structural inequalities by privileging actors who already have greater access to resources and formal markets (Aworka et al., 2022).
Evidence from the reviewed studies also indicates that participatory approaches to digital tool development were rarely employed. Most machine learning and remote sensing applications were developed in laboratory and research settings, with limited engagement of end users during problem definition, model development and validation (Khorramifar et al., 2023). Where participatory elements were included, studies reported improved relevance of decision-support outputs and greater alignment with local production and market realities. However, such approaches were the exception rather than the norm, highlighting a disconnect between technical innovation and inclusive value chain governance.
Taken together, the reviewed evidence indicates that the exclusion of gender-responsive and participatory design principles represents a critical limitation of current data-driven approaches to addressing sweet potato oversupply and post-harvest losses. While digital tools demonstrate strong technical potential to improve forecasting, coordination, and efficiency, their livelihood impacts remain unevenly distributed (Przybył et al., 2022). The findings therefore indicate that data-driven interventions, if developed without explicit attention to gender and social inclusion, may inadvertently reproduce or exacerbate existing disparities within smallholder food systems rather than contribute to equitable transformation (Tang et al., 2025).
Co-designing data-driven theoretical frameworks for inclusive food systems
A smaller subset of the reviewed literature (13%; n = 26) focused explicitly on the development of theoretical frameworks that integrate data-driven tools within broader food system and value chain governance structures. These studies moved beyond isolated empirical modelling to propose structured frameworks that link machine learning applications, simulation models, and policy-relevant decision-support systems to address systemic inefficiencies in agricultural value chains, including post-harvest losses and market volatility (Shoffe and Johnson, 2024).
Across these studies, data-driven frameworks were designed to operate at multiple nodes of the value chain, combining production-level analytics for example yield forecasting and spatial mapping, with post-harvest optimization tools such as storage simulations and loss-reduction scenarios (Wang and Hsu, 2024). In addition to market-oriented intelligence such as demand forecasting and supply–demand matching. Simulation-based approaches were frequently used to test alternative policy and investment scenarios. This approach allowed stakeholders to assess the potential impacts of storage infrastructure, processing capacity and market coordination mechanisms before implementation.
Importantly, the reviewed frameworks emphasized participatory governance as a core design principle. Rather than treating data analytics as a purely technical solution, these studies embedded data-driven tools within co-design processes involving farmers, extension services, processors, and policymakers (Coussement et al., 2024). This participatory orientation was particularly highlighted in relation to underutilized and semi-commercial crops that are frequently challenged by informal markets, seasonal abundance, and limited institutional support. The frameworks therefore positioned data science as an enabling layer that supports collective decision-making, adaptive planning, and inclusive value chain transformation, rather than as a standalone technological intervention.
Implications for policy, practice, and future research
The systematic review showed three key implications. First, data science must be applied across the entire sweet potato value chain rather than concentrated at the production stage only. Integrated dashboards linking production forecasts, market demand, and processing capacity could mitigate oversupply and reduce losses. Second, gender inclusion should be treated as a core design principle rather than an auxiliary objective. Embedding gender-disaggregated indicators and participatory co-design mechanisms can improve both equity and system performance. Third, the limited number of conceptual and theoretical studies (13%) shows a need for validated frameworks that integrate digital tools, governance, and social inclusion within underutilized crop systems. The systematic review also indicates some areas of further research. These areas include empirical evaluation of data-driven interventions that target post-harvest coordination and small-scale processing at scale. For these interventions to be successful, collaboration among data-scientists, agronomists, social scientists and policy makers must be emphasized to develop context-sensitive solutions.
Study limitations
This systematic review synthesized emerging evidence on data science applications in sweet potato value chains and livelihoods, yet several limitations were noted. The analysis was constrained by the small number of studies addressing gender and downstream value chain nodes, and the predominance of cross-sectional and modelling studies limited causal inference on livelihood impacts. The review relied primarily on Google Scholar and Science Direct, potentially omitting relevant studies from other databases or grey literature. Only 26 articles published in English within the past 10 years were included, excluding earlier and non-English publications that may offer valuable insights.
Data science remains a novel area in Malawi, resulting in limited locally relevant research. Additionally, only open-access articles were considered due to funding constraints, which may have excluded important paid-access studies. These limitations restrict the generalizability of findings beyond the reviewed contexts, hence the need for expanded, inclusive, and context-sensitive research to strengthen the evidence base for data-driven interventions in agricultural value chains.
Conclusion
This systematic review showed that sweet potato production remains a critical pillar for food and nutrition security and rural livelihoods in Malawi and across sub-Saharan Africa. However, its full potential continues to be challenged by continuous challenges. These challenges include low productivity, limited access to improved planting materials, postharvest losses, and weak coordination across the value chain. These significant structural limitations contribute to recurrent oversupply and inefficiencies that affect both income generation and nutritional outcomes for smallholder farmers. The systematic review further concludes that data-driven tools especially machine learning, hyperspectral imaging, and remote sensing offer significant opportunities to address these constraints. While the technical advancements of these digital approaches are advancing rapidly, their practical application within smallholder-dominated contexts remains limited and uneven. This shows a need for context-sensitive scaling strategies.
The review also concludes that gender and social inclusion are insufficiently integrated into current data-driven sweet potato interventions. Additionally, only a small body of research has progressed beyond empirical modelling to develop integrated frameworks that embed data analytics within broader governance, policy, and food system structures. Addressing these gaps will be essential for ensuring that data-enabled innovations translate into inclusive, sustainable, and nutrition-sensitive livelihood improvements in Malawi and similar settings.
Policy and contextual implications
The findings of this systematic review point to clear policy and practice implications for strengthening sweet potato value chains and improving livelihoods in Malawi through more coordinated, evidence-informed approaches. While digital and data-driven tools show growing potential to enhance production planning, postharvest management, and market coordination, their effectiveness depends on how well they are embedded within existing agricultural, ICT, and governance frameworks. Integrating decision-support mechanisms such as production forecasting tools, market information systems, and early-warning platforms into national agricultural strategies, including the National Agriculture Policy (NAP), National Agriculture Investment Plan (NAIP) and associated investment frameworks, could help reduce recurrent inefficiencies linked to oversupply, postharvest losses, and weak market integration.
The review also shows the importance of institutional coordination and partnerships in translating emerging analytical tools into practical outcomes for smallholder farmers. Public–private partnerships, supported by development partners, can play a catalytic role in development of decision-support systems that are accessible, affordable, and responsive to local conditions. At the same time, the establishment of clear data governance arrangements is essential to safeguard farmer interests, promote transparency, and ensure that data generated through agricultural systems are used equitably.
Finally, the review shows opportunities for co-designing gender-sensitive integrated frameworks that support underutilized and climate-resilient crops such as sweet potato within broader food system transformation efforts. Multi-stakeholder platforms that bring together researchers, policymakers, extension services, and agribusiness actors enhance practical solutions to postharvest losses, value addition, and market expansion. As indicated in the Malawi Growth and Development Strategy III, aligning these efforts with national development priorities and regional food system initiatives can support a more coherent and inclusive pathway toward agricultural diversification, resilience, and livelihood improvement in Malawi, while offering transferable insights for similar contexts across sub-Saharan Africa.
Final remarks
The reviewed evidence shows that data-driven approaches including predictive analytics, remote sensing, and digital decision-support systems strengthen the sweet potato value chain. By systematically synthesizing evidence at the intersection of oversupply dynamics, data-informed interventions, and livelihood outcomes, this review fills an important gap in the literature. The findings emphasize that technological solutions alone are insufficient. The limited incorporation of gender-responsive and participatory designs constrains equitable impact. To achieve sustainable transformation, data-informed innovations must be embedded within inclusive governance, institutional, and policy frameworks that enhance resilience, expand opportunities for women and youth, and reposition sweet potato oversupply as a pathway to inclusive and adaptive food systems in resource-constrained contexts.
Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
The authors would like to acknowledge the technical support and mentorship they received from various departments such as; Food Science and Technology; Food technology; and extension at the Lilongwe University of Agriculture and Natural Resources (LUANAR). Research institutions including International Center for Tropical Agriculture (CIAT) and Malawi University of Science and Technology (MUST). This systematic review paper was made possible because of your priceless guidance, efforts and insights. We also thank the reviewers of the manuscript for their unwavering support and constructive feedback to improve and clarify it until publication.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that Generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
ReferencesAbidinS.ReddyK.ZhangC. (2015). Intensity of price and volatility spillover effects in Asia-Pacific Basin equity markets. Australas. Account. Bus. Finance J.8, 3–18. doi: 10.14453/aabfj.v8i5.2AfzalN.AfionisS.StringerL. C.FavrettoN.SakaiM.SakaiP. (2021). Benefits and trade-offs of smallholder sweet potato cultivation as a pathway toward achieving the sustainable development goals. Sustainability13:552. doi: 10.3390/su13020552AhmedM. T.VillordonA.KamruzzamanM. (2025). Hyperspectral imaging and explainable deep-learning for non-destructive quality prediction of sweetpotato. Postharvest Biol. Technol.222:113379. doi: 10.1016/j.postharvbio.2024.113379AlamuE. O.AdesokanM.AsfawA.Maziya-DixonB. (2020). Effect of sample preparation methods on the prediction performances of near infrared reflectance spectroscopy for quality traits of fresh yam (Dioscorea spp.). Appl. Sci.10:6035. doi: 10.3390/app10176035AworkaR.CedricL. S.AdoniW. Y. H.ZoueuJ. T.MutomboF. K.KimpoloC. L. M.. (2022). Agricultural decision system based on advanced machine learning models for yield prediction: case of east African countries. Smart Agric. Technol.2:100048. doi: 10.1016/j.atech.2022.100048Carbajal-CarrascoM.JonesD.WilliamsC.NelsonN. (2024). In-season sweetpotato yield forecasting using multitemporal remote sensing environmental observations and machine learning. ESS Open Arch. doi: 10.22541/essoar.171415899.94920111/v1Cartabiano-LeiteC. E.PorcuO. M.de CasasA. F. (2020). Sweet potato (Ipomoea batatas L. lam) nutritional potential and social relevance: a review. History11, 23–40. doi: 10.9790/9622-1006082340CoussementK.AbedinM. Z.KrausM.MaldonadoS.TopuzK. (2024). Explainable AI for enhanced decision-making. Decis. Support. Syst.184:114276. doi: 10.1016/j.dss.2024.114276DiopP. M.OshiroN.NakamuraM.TakamotoJ.NakamuraY. (2023). Design of machine learning solutions to post-harvest classification of vegetal species. AgriEngineering5, 1005–1019. doi: 10.3390/agriengineering5020063DubeyS.SinghR.SinghS. P.MishraA. H.SinghN. V. (2020). A brief study of value chain and supply chain. Agriculture development and economic transformation in global scenario, 177-183Available online at: https://www.researchgate.net/publication/344374264_A_Brief_Study_Of_Value_Chain_And_Supply_ChainGelayeY. (2024). Sweet potato: a versatile solution for nutritional challenges in Ethiopia. Technol. Agron.4. doi: 10.48130/tia-0024-0011GereffiG.Fernandez-StarkK. (2011). Global value chain analysis: A primer. Center on globalization, Governance & Competitiveness (CGGC), vol. 33. Durham, NC: Duke University, 54–76.GholamiS.KnippenbergE.CampbellJ.AndriantsimbaD.KamleA.ParthasarathyP.. (2022). Food security analysis and forecasting: a machine learning case study in southern Malawi. Data Policy4:e33. doi: 10.1017/dap.2022.25IslamS. (2024). Sweetpotatoes [Ipomoea batatas (L.) lam]: the super food of the next century? An intensive review on their potential as a sustainable and versatile food source for future generations. CyTA – J Food22:2397553. doi: 10.1080/19476337.2024.2397553KapalasaE.MwenyeO.JogoW.ParkerM.DemoP. (2022). Potato value chain analysis report for Malawi. Lima, Peru.KawayeF. P.HutchinsonM. F. (2020). “Maize, cassava, and sweet potato yield on monthly climate in Malawi” in African handbook of climate change adaptation (Cham: Springer International Publishing), 1–21.KhorramifarA.RasekhM.KaramiH.LozanoJ.GancarzM.ŁazukaE.. (2023). Determining the shelf life and quality changes of potatoes (Solanum tuberosum) during storage using electronic nose and machine learning. PLoS One18:e0284612. doi: 10.1371/journal.pone.0284612, 37115737KpienbaarehD.SunX.WangJ.LuginaahI.Bezner KerrR.LupafyaE.. (2021). Crop type and land cover mapping in northern Malawi using the integration of sentinel-1, sentinel-2, and planetscope satellite data. Remote Sens13:700. doi: 10.3390/rs13040700KurekJ.NiedbałaG.WojciechowskiT.ŚwiderskiB.AntoniukI.PiekutowskaM.. (2023). Prediction of potato (Solanum tuberosum L.) yield based on machine learning methods. Agriculture13:2259. doi: 10.3390/agriculture13122259LiuH.HuntS.YenchoG. C.PecotaK. V.MieropR.WilliamsC. M.. (2024). Predicting sweetpotato traits using machine learning: impact of environmental and agronomic factors on shape and size. Comput. Electron. Agric.225:109215. doi: 10.1016/j.compag.2024.109215LongweK.AkiniwaleG.MwenyeJ. O. (2023). Effects of intercropping sweetpotato (Ipomea. Batatas) with pigeonpea on productivity and major pest of sweetpotato in Malawi. FARA Res. Rep.7, 800–809. doi: 10.59101/frr072362LowJ. W.MwangaR. O.AndradeM.CareyE.BallA. M. (2017). Tackling vitamin a deficiency with biofortified sweetpotato in sub-Saharan Africa. Glob. Food Sec.14, 23–30. doi: 10.1016/j.gfs.2017.01.004, 28989861MotsaN. M.ModiA. T.MabhaudhiT. (2015). Sweet potato (Ipomoea batatas L.) as a drought tolerant and food security crop. S. Afr. J. Sci.111, 1–8. doi: 10.17159/sajs.2015/20140252MungaiL. M.MessinaJ. P.ZuluL. C.QiJ.SnappS. (2022). Modeling spatiotemporal patterns of land use/land cover change in Central Malawi using a neural network model. Remote Sens14:3477. doi: 10.3390/rs14143477MunyakaJ. C. B.GallayO.ChenalJ.MutandwaE.SalgadoX.PindayiT.. (2025). Bridging the gender gap in climate-resilient sweet potato farming: a case study from Goromonzi District, Zimbabwe. Systems13:135. doi: 10.3390/systems13020135NowakowskiA.Del RossoM. P.ZacharP.SpillerD.GabaraG.BarrettaD.. (2024). Transfer learning in earth observation data analysis: a review. IEEE Geosci. Remote Sens. Mag. doi: 10.1109/MGRS.2024.3494673PageM. J.McKenzieJ. E.BossuytP. M.BoutronI.HoffmannT. C.MulrowC. D.. (2021). The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ372. doi: 10.1136/bmj.n71PrzybyłK.AdamskiF.WawrzyniakJ.Gawrysiak-WitulskaM.StangierskiJ.KmiecikD. (2022). Machine and deep learning in the evaluation of selected qualitative characteristics of sweet potatoes obtained under different convective drying conditions. Appl. Sci.12:7840. doi: 10.3390/app12157840RamírezD. A.GrünebergW.I AndradeM.De BoeckB.LoayzaH.S MakundeG.. (2023). Phenotyping of productivity and resilience in sweetpotato under water stress through UAV-based multispectral and thermal imagery in Mozambique. J. Agron. Crop Sci.209, 41–55. doi: 10.1111/jac.12565ShoffeY.JohnsonL. K. (2024). Opportunities for prediction models to reduce food loss and waste in the postharvest chain of horticultural crops. Sustainability16:7803. doi: 10.3390/su16177803SuW. H.XueH. (2021). Imaging spectroscopy and machine learning for intelligent determination of potato and sweet potato quality. Foods10:2146. doi: 10.3390/foods10092146, 34574253TangC.MoX.MaZ.WangZ. (2025). Machine learning-enhanced near-infrared spectroscopy for high-throughput phenotyping of sweetpotato sugars across raw and cooked states. J. Agric. Food Res.21:101934. doi: 10.1016/j.jafr.2025.101934Van VugtD.FrankeA. C. (2022). Exploring the yield gap of orange-fleshed sweet potato varieties on smallholder farmers’ fields in Malawi. Field Crop Res.221, 245–256. doi: 10.1016/j.fcr.2017.11.028WangL.HsuH. H. (2024). “The utilization of geographic information system in logistics distribution” in Proceedings of the 2024 2nd international conference on internet of things and cloud computing technology, 131–135. doi: 10.1145/3702879.3702902YadavN.SidanaN. (2023). Precision agriculture technologies: analysing the use of advanced technologies, such as drones, sensors, and GPS, in precision agriculture for optimizing resource management, crop monitoring, and yield prediction. J. Adv. Zool.44. doi: 10.53555/jaz.v44iS5.716ZamoraE. A. (2016). Value chain analysis: a brief review. Asian J. Innov. Policy5, 116–128. doi: 10.7545/ajip.2016.5.2.116
Edited by: Mjabuliseni Ngidi, University of KwaZulu-Natal, South Africa
Reviewed by: Mario Alejandro Hernández Chontal, Universidad Veracruzana, Mexico