Recurrent neural networks enable design of multifunctional synthetic human gut microbiome dynamics

Decoding the role of the arginine dihydrolase pathway in shaping human gut community assembly and health-relevant metabolites

The arginine dihydrolase pathway (arc operon) provides a metabolic niche by transforming arginine into metabolic byproducts. We investigate the role of the arc operon in probiotic Escherichia coli Nissle 1917 on human gut community assembly and health-relevant metabolite profiles. By stabilizing environmental pH, the arc operon reduces variability in community composition in response to pH perturbations and frequently enhances butyrate production in synthetic communities. We use a tailored machine learning model for microbiomes to predict community assembly in response to variation in initial media pH and arc operon activity. This model uncovers the pH- and arc operon-dependent interactions shaping community assembly. Human gut species display altered colonization dynamics in response to the arc operon in the murine gut. In sum, our framework to quantify the contribution of a specific pathway to microbial community assembly and metabolite production can reveal new engineering strategies. A record of this paper's transparent peer review process is included in the supplemental information.

Distinguishing critical microbial community shifts from normal temporal variability in human and environmental ecosystems

Differentiating significant microbial community changes from normal fluctuations is vital for understanding microbial dynamics in human and environmental ecosystems. This knowledge could enable early warning systems to monitor critical changes affecting human or environmental health. We applied 16S rRNA gene sequencing and time-series analysis to model bacterial abundance trajectories in human gut and wastewater microbiomes. We evaluated various model architectures using datasets from two human studies and five wastewater settings. Long short-term memory (LSTM) models consistently outperformed other models in predicting bacterial abundances and detecting outliers, as measured by multiple metrics. Prediction intervals for each genus allowed us to identify significant changes and signaling shifts in community states. This study proposes a machine learning model capable of monitoring microbial communities and providing insights into their responses to internal and external factors in medical and environmental settings.

Artificial-intelligence-driven Innovations in Mechanistic Computational Modeling and Digital Twins for Biomedical Applications

Understanding of complex biological systems remains a significant challenge due to their high dimensionality, nonlinearity, and context-specific behavior. Artificial intelligence (AI) and mechanistic modeling are becoming essential tools for studying such complex systems. Mechanistic modeling can facilitate the construction of simulatable models that are interpretable but often struggle with scalability and parameters estimation. AI can integrate multi-omics data to create predictive models, but it lacks interpretability. The gap between these two modeling methods limits our ability to develop comprehensive and predictive models for biomedical applications. This article reviews the most recent advancements in the integration of AI and mechanistic modeling to fill this gap. Recently, with omics availability, AI has led to new discoveries in mechanistic computational modeling. The mechanistic models can also help in getting insight into the mechanism for prediction made by AI models. This integration is helpful in modeling complex systems, estimating the parameters that are hard to capture in experiments, and creating surrogate models to reduce computational costs because of expensive mechanistic model simulations. This article focuses on advancements in mechanistic computational models and AI models and their integration for scientific discoveries in biology, pharmacology, drug discovery and diseases. The mechanistic models with AI integration can facilitate biological discoveries to advance our understanding of disease mechanisms, drug development, and personalized medicine. The article also highlights the role of AI and mechanistic model integration in the development of more advanced models in the biomedical domain, such as medical digital twins and virtual patients for pharmacological discoveries.

Integrating microbial communities into algal biotechnology: a pathway to enhanced commercialization

Microalgae are increasingly recognized for their potential in wastewater treatment and the sustainable production of feedstock for fuel, feed, food, and other bioproducts. Like conventional agricultural systems, algal cultivation involves complex microbial communities. However, despite their pivotal role in cultivation outcomes, especially at the commodity-scale, the critical interactions between microalgae and their microbiomes are often overlooked. Here we synthesize current knowledge on the taxonomic diversity, ecological roles, and biotechnological potential of algal microbiomes, with a focus on their interactions with algal hosts through nutrient exchange, growth modulation, pathogen defense, and environmental conditioning. We also examine how environmental factors such as nutrient availability, salinity, and temperature influence these interactions. Advances in microbiome engineering, including synthetic biology and ecological approaches, offer opportunities to enhance beneficial algal-microbiome interactions, thereby improving growth, resilience, and yield. These advancements could lead to more sustainable and economically viable microalgae cultivation, with far-reaching implications for environmental management and biotechnological innovation. By addressing key economic and environmental barriers, microbiome engineering holds transformative potential to revolutionize large-scale algae cultivation and provide sustainable solutions to global challenges.

Dual graph-embedded fusion network for predicting potential microbe-disease associations with sequence learning

Recent studies indicate that microorganisms are crucial for maintaining human health. Dysbiosis, or an imbalance in these microbial communities, is strongly linked to a variety of human diseases. Therefore, understanding the impact of microbes on disease is essential. The DuGEL model leverages the strengths of graph convolutional neural network (GCN) and graph attention network (GAT), ensuring that both local and global relationships within the microbe-disease association network are captured. The integration of the Long Short-Term Memory Network (LSTM) further enhances the model's ability to understand sequential dependencies in the feature representations. This comprehensive approach allows DuGEL to achieve a high level of accuracy in predicting potential microbe-disease associations, making it a valuable tool for biomedical research and the discovery of new therapeutic targets. By combining advanced graph-based and sequence-based learning techniques, DuGEL addresses the limitations of existing methods and provides a robust framework for the prediction of microbe-disease associations. To evaluate the performance of DuGEL, we conducted comprehensive comparative experiments and case studies based on two databases, HMDAD, and Disbiome to demonstrate that DuGEL can effectively predict potential microbe-disease associations.

The architecture of theory and data in microbiome design: towards an S-matrix for microbiomes

Designing microbiomes for applications in health, bioengineering, and sustainability is intrinsically linked to a fundamental theoretical understanding of the rules governing microbial community assembly. Microbial ecologists have used a range of mathematical models to understand, predict, and control microbiomes, ranging from mechanistic models, putting microbial populations and their interactions as the focus, to purely statistical approaches, searching for patterns in empirical and experimental data. We review the success and limitations of these modeling approaches when designing novel microbiomes, especially when guided by (inevitably) incomplete experimental data. Although successful at predicting generic patterns of community assembly, mechanistic and phenomenological models tend to fall short of the precision needed to design and implement specific functionality in a microbiome. We argue that to effectively design microbiomes with optimal functions in diverse environments, ecologists should combine data-driven techniques with mechanistic models - a middle, third way for using theory to inform design.

Deep learning in microbiome analysis: a comprehensive review of neural network models

Microbiome research, the study of microbial communities in diverse environments, has seen significant advances due to the integration of deep learning (DL) methods. These computational techniques have become essential for addressing the inherent complexity and high-dimensionality of microbiome data, which consist of different types of omics datasets. Deep learning algorithms have shown remarkable capabilities in pattern recognition, feature extraction, and predictive modeling, enabling researchers to uncover hidden relationships within microbial ecosystems. By automating the detection of functional genes, microbial interactions, and host-microbiome dynamics, DL methods offer unprecedented precision in understanding microbiome composition and its impact on health, disease, and the environment. However, despite their potential, deep learning approaches face significant challenges in microbiome research. Additionally, the biological variability in microbiome datasets requires tailored approaches to ensure robust and generalizable outcomes. As microbiome research continues to generate vast and complex datasets, addressing these challenges will be crucial for advancing microbiological insights and translating them into practical applications with DL. This review provides an overview of different deep learning models in microbiome research, discussing their strengths, practical uses, and implications for future studies. We examine how these models are being applied to solve key problems and highlight potential pathways to overcome current limitations, emphasizing the transformative impact DL could have on the field moving forward.

Effects of data transformation and model selection on feature importance in microbiome classification data

BACKGROUND

Accurate classification of host phenotypes from microbiome data is crucial for advancing microbiome-based therapies, with machine learning offering effective solutions. However, the complexity of the gut microbiome, data sparsity, compositionality, and population-specificity present significant challenges. Microbiome data transformations can alleviate some of the aforementioned challenges, but their usage in machine learning tasks has largely been unexplored.

RESULTS

Our analysis of over 8500 samples from 24 shotgun metagenomic datasets showed that it is possible to classify healthy and diseased individuals using microbiome data with minimal dependence on the choice of algorithm or transformation. Presence-absence transformations performed comparably to abundance-based transformations, and only a small subset of predictors is necessary for accurate classification. However, while different transformations resulted in comparable classification performance, the most important features varied significantly, which highlights the need to reevaluate machine learning-based biomarker detection.

CONCLUSIONS

Microbiome data transformations can significantly influence feature selection but have a limited effect on classification accuracy. Our findings suggest that while classification is robust across different transformations, the variation in feature selection necessitates caution when using machine learning for biomarker identification. This research provides valuable insights for applying machine learning to microbiome data and identifies important directions for future work.

Assembly Graph as the Rosetta Stone of Ecological Assembly

Ecological assembly-the process of ecological community formation through species introductions-has recently seen exciting theoretical advancements across dynamical, informational, and probabilistic approaches. However, these theories often remain inaccessible to non-theoreticians, and they lack a unifying lens. Here, I introduce the assembly graph as an integrative tool to connect these emerging theories. The assembly graph visually represents assembly dynamics, where nodes symbolise species combinations and edges represent transitions driven by species introductions. Through the lens of assembly graphs, I review how ecological processes reduce uncertainty in random species arrivals (informational approach), identify graphical properties that guarantee species coexistence and examine how the class of dynamical models constrain the topology of assembly graphs (dynamical approach), and quantify transition probabilities with incomplete information (probabilistic approach). To facilitate empirical testing, I also review methods to decompose complex assembly graphs into smaller, measurable components, as well as computational tools for deriving empirical assembly graphs. In sum, this math-light review of theoretical progress aims to catalyse empirical research towards a predictive understanding of ecological assembly.

Predicting and Prioritising Community Assembly: Learning Outcomes via Experiments

Community assembly provides the foundation for applications in biodiversity conservation, climate change, invasion, restoration and synthetic ecology. However, predicting and prioritising assembly outcomes remains difficult. We address this challenge via a mechanism-free approach useful when little data or knowledge exist (LOVE; Learning Outcomes Via Experiments). We carry out assembly experiments ('actions', here, random combinations of species additions) potentially in multiple environments, wait, and measure abundance outcomes. We then train a model to predict outcomes of novel actions or prioritise actions that would yield the most desirable outcomes. Across 10 single- and multi-environment datasets, when trained on 89 randomly selected actions, LOVE predicts outcomes with 0.5%-3.4% mean error, and prioritises actions for maximising richness, maximising abundance, or removing unwanted species, with 94%-99% mean true positive rate and 10%-84% mean true negative rate across tasks. LOVE complements existing mechanism-first approaches for community ecology and may help address numerous applied challenges.

Elucidating human gut microbiota interactions that robustly inhibit diverse Clostridioides difficile strains across different nutrient landscapes

The human gut pathogen Clostridioides difficile displays substantial inter-strain genetic variability and confronts a changeable nutrient landscape in the gut. We examined how human gut microbiota inter-species interactions influence the growth and toxin production of various C. difficile strains across different nutrient environments. Negative interactions influencing C. difficile growth are prevalent in an environment containing a single highly accessible resource and sparse in an environment containing C. difficile-preferred carbohydrates. C. difficile toxin production displays significant community-context dependent variation and does not trend with growth-mediated inter-species interactions. C. difficile strains exhibit differences in interactions with Clostridium scindens and the ability to compete for proline. Further, C. difficile shows substantial differences in transcriptional profiles in co-culture with C. scindens or Clostridium hiranonis. C. difficile exhibits massive alterations in metabolism and other cellular processes in co-culture with C. hiranonis, reflecting their similar metabolic niches. C. hiranonis uniquely inhibits the growth and toxin production of diverse C. difficile strains across different nutrient environments and robustly ameliorates disease severity in mice. In sum, understanding the impact of C. difficile strain variability and nutrient environments on inter-species interactions could help improve the effectiveness of anti-C. difficile strategies.

Human gut microbiota interactions shape the long-term growth dynamics and evolutionary adaptations of Clostridioides difficile

Clostridioides difficile can transiently or persistently colonize the human gut, posing a risk factor for infections. This colonization is influenced by complex molecular and ecological interactions with human gut microbiota. By investigating C. difficile dynamics in human gut communities over hundreds of generations, we show patterns of stable coexistence, instability, or competitive exclusion. Lowering carbohydrate concentration shifted a community containing C. difficile and the prevalent human gut symbiont Phocaeicola vulgatus from competitive exclusion to coexistence, facilitated by increased cross-feeding. In this environment, C. difficile adapted via single-point mutations in key metabolic genes, altering its metabolic niche from proline to glucose utilization. These metabolic changes substantially impacted inter-species interactions and reduced disease severity in the mammalian gut. In sum, human gut microbiota interactions are crucial in shaping the long-term growth dynamics and evolutionary adaptations of C. difficile, offering key insights for developing anti-C. difficile strategies.

Elucidating human gut microbiota interactions that robustly inhibit diverse Clostridioides difficile strains across different nutrient landscapes

The human gut pathogen Clostridioides difficile displays extreme genetic variability and confronts a changeable nutrient landscape in the gut. We mapped gut microbiota inter-species interactions impacting the growth and toxin production of diverse C. difficile strains in different nutrient environments. Although negative interactions impacting C. difficile are prevalent in environments promoting resource competition, they are sparse in an environment containing C. difficile-preferred carbohydrates. C. difficile strains display differences in interactions with Clostridium scindens and the ability to compete for proline. C. difficile toxin production displays substantial community-context dependent variation and does not trend with growth-mediated inter-species interactions. C. difficile shows substantial differences in transcriptional profiles in the presence of the closely related species C. hiranonis or C. scindens. In co-culture with C. hiranonis, C. difficile exhibits massive alterations in metabolism and other cellular processes, consistent with their high metabolic overlap. Further, Clostridium hiranonis inhibits the growth and toxin production of diverse C. difficile strains across different nutrient environments and ameliorates the disease severity of a C. difficile challenge in a murine model. In sum, strain-level variability and nutrient environments are major variables shaping gut microbiota interactions with C. difficile.

Shaping human gut community assembly and butyrate production by controlling the arginine dihydrolase pathway

The arginine dihydrolase pathway (arc operon) present in a subset of diverse human gut species enables arginine catabolism. This specialized metabolic pathway can alter environmental pH and nitrogen availability, which in turn could shape gut microbiota inter-species interactions. By exploiting synthetic control of gene expression, we investigated the role of the arc operon in probiotic Escherichia coli Nissle 1917 on human gut community assembly and health-relevant metabolite profiles in vitro and in the murine gut. By stabilizing environmental pH, the arc operon reduced variability in community composition across different initial pH perturbations. The abundance of butyrate producing bacteria were altered in response to arc operon activity and butyrate production was enhanced in a physiologically relevant pH range. While the presence of the arc operon altered community dynamics, it did not impact production of short chain fatty acids. Dynamic computational modeling of pH-mediated interactions reveals the quantitative contribution of this mechanism to community assembly. In sum, our framework to quantify the contribution of molecular pathways and mechanism modalities on microbial community dynamics and functions could be applied more broadly.

Artificial intelligence-driven microbiome data analysis for estimation of postmortem interval and crime location

Microbial communities, demonstrating dynamic changes in cadavers and the surroundings, provide invaluable insights for forensic investigations. Conventional methodologies for microbiome sequencing data analysis face obstacles due to subjectivity and inefficiency. Artificial Intelligence (AI) presents an efficient and accurate tool, with the ability to autonomously process and analyze high-throughput data, and assimilate multi-omics data, encompassing metagenomics, transcriptomics, and proteomics. This facilitates accurate and efficient estimation of the postmortem interval (PMI), detection of crime location, and elucidation of microbial functionalities. This review presents an overview of microorganisms from cadavers and crime scenes, emphasizes the importance of microbiome, and summarizes the application of AI in high-throughput microbiome data processing in forensic microbiology.

Control points for design of taxonomic composition in synthetic human gut communities

Microbial communities offer vast potential across numerous sectors but remain challenging to systematically control. We develop a two-stage approach to guide the taxonomic composition of synthetic microbiomes by precisely manipulating media components and initial species abundances. By combining high-throughput experiments and computational modeling, we demonstrate the ability to predict and design the diversity of a 10-member synthetic human gut community. We reveal that critical environmental factors governing monoculture growth can be leveraged to steer microbial communities to desired states. Furthermore, systematically varied initial abundances drive variation in community assembly and enable inference of pairwise inter-species interactions via a dynamic ecological model. These interactions are overall consistent with conditioned media experiments, demonstrating that specific perturbations to a high-richness community can provide rich information for building dynamic ecological models. This model is subsequently used to design low-richness communities that display low or high temporal taxonomic variability over an extended period. A record of this paper's transparent peer review process is included in the supplemental information.

Sparsity of higher-order landscape interactions enables learning and prediction for microbiomes

Microbiome engineering offers the potential to leverage microbial communities to improve outcomes in human health, agriculture, and climate. To translate this potential into reality, it is crucial to reliably predict community composition and function. But a brute force approach to cataloging community function is hindered by the combinatorial explosion in the number of ways we can combine microbial species. An alternative is to parameterize microbial community outcomes using simplified, mechanistic models, and then extrapolate these models beyond where we have sampled. But these approaches remain data-hungry, as well as requiring an a priori specification of what kinds of mechanisms are included and which are omitted. Here, we resolve both issues by introducing a mechanism-agnostic approach to predicting microbial community compositions and functions using limited data. The critical step is the identification of a sparse representation of the community landscape. We then leverage this sparsity to predict community compositions and functions, drawing from techniques in compressive sensing. We validate this approach on in silico community data, generated from a theoretical model. By sampling just [Formula: see text]1% of all possible communities, we accurately predict community compositions out of sample. We then demonstrate the real-world application of our approach by applying it to four experimental datasets and showing that we can recover interpretable, accurate predictions on composition and community function from highly limited data.

Integrating a tailored recurrent neural network with Bayesian experimental design to optimize microbial community functions

Microbiomes interact dynamically with their environment to perform exploitable functions such as production of valuable metabolites and degradation of toxic metabolites for a wide range of applications in human health, agriculture, and environmental cleanup. Developing computational models to predict the key bacterial species and environmental factors to build and optimize such functions are crucial to accelerate microbial community engineering. However, there is an unknown web of interactions that determine the highly complex and dynamic behavior of these systems, which precludes the development of models based on known mechanisms. By contrast, entirely data-driven machine learning models can produce physically unrealistic predictions and often require significant amounts of experimental data to learn system behavior. We develop a physically-constrained recurrent neural network that preserves model flexibility but is constrained to produce physically consistent predictions and show that it can outperform existing machine learning methods in the prediction of certain experimentally measured species abundance and metabolite concentrations. Further, we present a closed-loop, Bayesian experimental design algorithm to guide data collection by selecting experimental conditions that simultaneously maximize information gain and target microbial community functions. Using a bioreactor case study, we demonstrate how the proposed framework can be used to efficiently navigate a large design space to identify optimal operating conditions. The proposed methodology offers a flexible machine learning approach specifically tailored to optimize microbiome target functions through the sequential design of informative experiments that seek to explore and exploit community functions.

Deep Neural Networks for Predicting Single-Cell Responses and Probability Landscapes

Engineering biology relies on the accurate prediction of cell responses. However, making these predictions is challenging for a variety of reasons, including the stochasticity of biochemical reactions, variability between cells, and incomplete information about underlying biological processes. Machine learning methods, which can model diverse input-output relationships without requiring a priori mechanistic knowledge, are an ideal tool for this task. For example, such approaches can be used to predict gene expression dynamics given time-series data of past expression history. To explore this application, we computationally simulated single-cell responses, incorporating different sources of noise and alternative genetic circuit designs. We showed that deep neural networks trained on these simulated data were able to correctly infer the underlying dynamics of a cell response even in the presence of measurement noise and stochasticity in the biochemical reactions. The training set size and the amount of past data provided as inputs both affected prediction quality, with cascaded genetic circuits that introduce delays requiring more past data. We also tested prediction performance on a bistable auto-activation circuit, finding that our initial method for predicting a single trajectory was fundamentally ill-suited for multimodal dynamics. To address this, we updated the network architecture to predict the entire distribution of future states, showing it could accurately predict bimodal expression distributions. Overall, these methods can be readily applied to the diverse prediction tasks necessary to predict and control a variety of biological circuits, a key aspect of many synthetic biology applications.

Navigation between initial and desired community states using shortcuts

Ecological management problems often involve navigating from an initial to a desired community state. We ask whether navigation without brute-force additions and deletions of species is possible via: adding/deleting a small number of individuals of a species, changing the environment, and waiting. Navigation can yield direct paths (single sequence of actions) or shortcut paths (multiple sequences of actions with lower cost than a direct path). We ask (1) when is non-brute-force navigation possible?; (2) do shortcuts exist and what are their properties?; and (3) what heuristics predict shortcut existence? Using a state diagram framework applied to several empirical datasets, we show that (1) non-brute-force navigation is only possible between some state pairs, (2) shortcuts exist between many state pairs; and (3) changes in abundance and richness are the strongest predictors of shortcut existence, independent of dataset and algorithm choices. State diagrams thus unveil hidden strategies for manipulating species coexistence and efficiently navigating between states.

The community-function landscape of microbial consortia

Quantitatively linking the composition and function of microbial communities is a major aspiration of microbial ecology. Microbial community functions emerge from a complex web of molecular interactions between cells, which give rise to population-level interactions among strains and species. Incorporating this complexity into predictive models is highly challenging. Inspired by a similar problem in genetics of predicting quantitative phenotypes from genotypes, an ecological community-function (or structure-function) landscape could be defined that maps community composition and function. In this piece, we present an overview of our current understanding of these community landscapes, their uses, limitations, and open questions. We argue that exploiting the parallels between both landscapes could bring powerful predictive methodologies from evolution and genetics into ecology, providing a boost to our ability to engineer and optimize microbial consortia.

Controlling the human microbiome

We coexist with a vast number of microbes that live in and on our bodies. Those microbes and their genes are collectively known as the human microbiome, which plays important roles in human physiology and diseases. We have acquired extensive knowledge of the organismal compositions and metabolic functions of the human microbiome. However, the ultimate proof of our understanding of the human microbiome is reflected in our ability to manipulate it for health benefits. To facilitate the rational design of microbiome-based therapies, there are many fundamental questions to be addressed at the systems level. Indeed, we need a deep understanding of the ecological dynamics associated with such a complex ecosystem before we rationally design control strategies. In light of this, this review discusses progress from various fields, e.g., community ecology, network science, and control theory, that are helping us make progress toward the ultimate goal of controlling the human microbiome.

Video frame prediction of microbial growth with a recurrent neural network

The recent explosion of interest and advances in machine learning technologies has opened the door to new analytical capabilities in microbiology. Using experimental data such as images or videos, machine learning, in particular deep learning with neural networks, can be harnessed to provide insights and predictions for microbial populations. This paper presents such an application in which a Recurrent Neural Network (RNN) was used to perform prediction of microbial growth for a population of two Pseudomonas aeruginosa mutants. The RNN was trained on videos that were acquired previously using fluorescence microscopy and microfluidics. Of the 20 frames that make up each video, 10 were used as inputs to the network which outputs a prediction for the next 10 frames of the video. The accuracy of the network was evaluated by comparing the predicted frames to the original frames, as well as population curves and the number and size of individual colonies extracted from these frames. Overall, the growth predictions are found to be accurate in metrics such as image comparison, colony size, and total population. Yet, limitations exist due to the scarcity of available and comparable data in the literature, indicating a need for more studies. Both the successes and challenges of our approach are discussed.

Carbohydrate complexity limits microbial growth and reduces the sensitivity of human gut communities to perturbations

Dietary fibre impacts the growth dynamics of human gut microbiota, yet we lack a detailed and quantitative understanding of how these nutrients shape microbial interaction networks and responses to perturbations. By building human gut communities coupled with computational modelling, we dissect the effects of fibres that vary in chemical complexity and each of their constituent sugars on community assembly and response to perturbations. We demonstrate that the degree of chemical complexity across different fibres limits microbial growth and the number of species that can utilize these nutrients. The prevalence of negative interspecies interactions is reduced in the presence of fibres compared with their constituent sugars. Carbohydrate chemical complexity enhances the reproducibility of community assembly and resistance of the community to invasion. We demonstrate that maximizing or minimizing carbohydrate competition between resident and invader species enhances resistance to invasion. In sum, the quantitative effects of carbohydrate chemical complexity on microbial interaction networks could be exploited to inform dietary and bacterial interventions to modulate community resistance to perturbations.

Evaluation of the Practical Effects of Environmental Measures in the Conservation of Architectural Heritage in Yan’an Based on Recurrent Neural Networks

Yan'an is one of the “two holy places” of the Chinese nation and the Chinese revolution and is one of the first cities of historical and cultural significance and an outstanding tourist city in China, as announced by the state council. The evaluation of the effectiveness of environmental conservation is one of the very important elements of the conservation of Yan'an's architectural heritage. However, the existing evaluation methods cannot provide new solutions for decision-making, the meaning of the comprehensive evaluation function is unclear, the naming clarity is low, there is less quantitative data and more qualitative components, and the results are not easily convincing. This paper proposes a method for evaluating the practical effects of environmental class measures in the conservation of Yan'an's architectural heritage based on recurrent neural networks. The recurrent neural network makes full use of the memory function in the network, considers the causal relationship of the actual effect, and efficiently evaluates the existing measures. In comparison with factor analysis and hierarchical analysis, this paper has greater applicability in evaluating the practical effects of environmental measures in the conservation of Yan'an's architectural heritage and is basically consistent with the results of the theoretical analysis. It provides a scientific basis for the construction and implementation of environmental measures for the architectural heritage of Yan'an.