The study was conducted between August 2025 and January 2026 by researchers in the Department of Pediatric Nursing at Bandırma Onyedi Eylül University, led by corresponding author Zübeyde Ezgi Erçelik. The team designed a cross-sectional, descriptive-correlational investigation and analyzed it using structural equation modeling, a statistical technique that allows researchers to test networks of hypothesized relationships among multiple variables simultaneously rather than examining each link in isolation. Before analysis began, the researchers screened the data for multivariate outliers and excluded five participants whose response patterns deviated substantially from the rest of the sample, leaving a final analytical cohort of 200 neonatal intensive care nurses.

The demographic profile of the sample reflects the realities of the nursing workforce in Türkiye’s neonatal units. Ninety-six percent of participants were female, with a mean age of 29.65 years and a standard deviation of 5.90. Most held a bachelor’s degree, at 62.5 percent, and 66.5 percent reported between one and five years of professional experience. What stands out most vividly, however, are the numbers on AI exposure: only 6.5 percent of these nurses had ever received any formal training in artificial intelligence, and just 13 percent reported using AI-based technologies in their clinical practice. In other words, the overwhelming majority of nurses caring for the country’s most vulnerable patients are navigating the arrival of AI with essentially no structured preparation.

Data collection relied on three instruments. A sociodemographic information form captured background characteristics, while the Artificial Intelligence Literacy Scale, abbreviated AILS, measured participants’ knowledge and competence across the dimensions of AI, from basic concepts to practical applications. The Medical Artificial Intelligence Readiness Scale, or MAIRS, assessed readiness across three subdomains: self-efficacy in using AI technologies, willingness to adopt them, and awareness of their benefits and limitations in medical settings. The researchers obtained permission from the authors of both scales before deployment, and the study received ethical approval from the Non-Interventional Clinical Research Ethics Committee of Bandırma Onyedi Eylul University in June 2025, with all procedures conducted in accordance with the Declaration of Helsinki.

The statistical architecture of the analysis deserves attention because it elevates the findings beyond simple correlation. Structural equation modeling combines factor analysis with path analysis, enabling the researchers to estimate how strongly one latent construct, AI literacy, predicts another, AI readiness, while accounting for measurement error in the scales themselves. The team evaluated the measurement models using established psychometric criteria, including average variance extracted and composite reliability to confirm that the scales measured their intended constructs consistently, and the heterotrait–monotrait ratio, or HTMT, to verify that literacy and readiness remained empirically distinct rather than collapsing into a single blurred variable. The study followed the STROBE guidelines for reporting observational research, a framework designed to strengthen the transparency and completeness of such studies.

The headline result is unambiguous. The model supported the study’s primary hypothesis, revealing that AI literacy significantly and positively predicted AI readiness, with a standardized path coefficient of 0.757 and a p-value below 0.001. In practical terms, a coefficient of this magnitude indicates that roughly three-quarters of a standard deviation increase in readiness accompanies each standard deviation increase in literacy, an exceptionally strong relationship in behavioral research. The literacy construct alone accounted for 57.3 percent of the variance in readiness scores, a proportion that social scientists rarely observe with a single predictor. Nurses with higher AI literacy scores did not merely report greater abstract enthusiasm; they reported greater self-efficacy and greater willingness regarding the integration of AI technologies into clinical decision-making processes.

Why does this matter so much for neonatal care specifically? AI tools are already entering the neonatal domain in forms ranging from predictive analytics for sepsis and deterioration risk to decision-support systems for respiratory management. These technologies promise faster and more accurate decision-making in units where minutes matter, but their benefits depend entirely on the humans operating them. A nurse who understands what an algorithm can and cannot do is better positioned to question its outputs, recognize when they conflict with bedside observation, and integrate them judiciously into care. The findings suggest that literacy is not a luxury but the gateway variable: improvements in AI literacy may directly enhance neonatal intensive care nurses’ readiness to weave AI into the decisions that determine newborn survival.

The near-total absence of prior training in the sample transforms the study from a descriptive snapshot into something closer to a warning. If 93.5 percent of neonatal nurses have received no AI education, and if literacy drives readiness this strongly, then the bottleneck is not nurse skepticism but institutional neglect. The authors argue that their empirical findings from Türkiye point toward a concrete intervention: structured AI literacy programs could raise readiness across the workforce, preparing nurses for technologies that are arriving whether or not hospitals plan for them. The researchers also note that participants with higher literacy reported greater self-efficacy and willingness, meaning that education would likely pay dividends not only in knowledge but in confidence and motivation.

There are limitations worth keeping in view. The cross-sectional design captures a single moment in time, so the modeling can demonstrate that literacy and readiness move together but cannot prove that boosting literacy causes readiness to rise; longitudinal and interventional studies would be needed to establish causality. The sample of 200 nurses, while adequate for the modeling technique, was drawn from a single national context, and the predominantly female, relatively young, bachelor’s-educated profile may not generalize to every workforce. Self-reported scales also depend on participants’ honest and accurate self-assessment. The authors received no specific grant from funding agencies in the public, commercial, or not-for-profit sectors, and they declare no competing interests.

Even with those caveats, the study lands at a pivotal moment. Hospitals worldwide are racing to deploy machine learning systems, yet the frontline clinicians who must live with those systems are rarely asked what they know or how prepared they feel. This research supplies the numbers: literacy is the strongest lever available, and it is currently almost untouched. For the nurses watching over newborns in intensive care, the message from the data is clear. The path to an AI-ready neonatal workforce runs directly through education, and the time to build that path is now, before the algorithms outpace the people expected to use them.

Subject of Research: Artificial intelligence literacy and readiness among neonatal nurses: a structural equation modeling study

Article Title: Artificial intelligence literacy and readiness among neonatal nurses: a structural equation modeling study

Article References: Erçelik, Z. E., Aydin Yilmaz, D., & Eryilmaz Taş, M. N. (2026). Artificial intelligence literacy and readiness among neonatal nurses: a structural equation modeling study. BMC Nursing. https://doi.org/10.1186/s12912-026-05337-w

Image Credits: AI Generated

DOI: 10.1186/s12912-026-05337-w

Keywords: Artificial, intelligence, literacy, readiness, neonatal, nurses, structural, equation, modeling, scientific research

The method, called Equation Discovery for Classification, or EDC, was developed by Guus Toussaint and Arno Knobbe of Leiden University. It extends a research tradition that has long flourished in regression: symbolic regression, the automated search for analytical formulas that fit numerical data. Theauthors of the study point out that equation discovery has historically been applied almost exclusively to problems where the target is a continuous number, such as recovering physical laws like the relationship between the pressure, volume, and temperature of a gas. Their contribution is to redirect that machinery toward binary classification, where the goal is to separate two classes cleanly and explainably.

The core idea is elegantly simple in concept. Rather than learning an opaque scoring function, EDC searches for a concise mathematical expression f(x) and a threshold theta, such that a data point is assigned to the positive class whenever f(x) meets or exceeds theta. For a linear equation, this is essentially the geometry underlying logistic regression or a linear support vector machine: a hyperplane slicing the feature space into two halves. What sets EDC apart is that the search is not confined to straight lines. By allowing nonlinear building blocks such as products of features and exponential terms, the algorithm can trace curved, interaction-driven boundaries while still producing an expression short enough for a domain expert to read, critique, and even correct by hand.

Technically, the framework rests on two pillars: a structured search and a dedicated numerical optimizer. The search space of candidate equations is defined by a configurable context-free grammar, a design choice inherited from classic work on declarative bias in equation discovery. The grammar constrains equations to sums of simple summands, including linear terms, products of two features, and exponential expressions, each parameterized by constants. Crucially, the grammar is redundancy-aware: constructions that would produce syntactically different but semantically identical equations are pruned in advance, since, for example, subtraction between summands is unnecessary when a constant coefficient can simply be negative. Traversing this space exhaustively is impossible for all but the smallest problems, so the algorithm employs beam search, iteratively refining the most promising candidate equations level by level while keeping only a fixed number of survivors at each stage.

Fitting the constants inside each candidate equation proved surprisingly subtle. Although every equation in the grammar is differentiable, gradient descent performed poorly in practice, largely because of the exponential terms that pepper the search space. The authors instead adopted a tailored hill-climbing procedure that first samples a large pool of random constant configurations, then concentrates its remaining budget on refining the best ones. In a systematic comparison against off-the-shelf optimizers from the SciPy library, including Powell, Cobyqa, Cobyla, Nelder-Mead, and stochastic gradient descent, this hill climber achieved the highest mean area under the ROC curve across one hundred randomly generated test problems. Perhaps most strikingly, a simple random-sampling baseline already reached a mean AUC of 0.9974 with a budget of one thousand evaluations, revealing that the inner optimization problem is more tractable than one might fear.

The experiments on synthetic data produced one of the study’s most intriguing findings. When Gaussian noise was injected into datasets whose generating decision boundaries were known, EDC did not merely match the original boundary; beyond a certain noise level it outperformed it. The explanation is a phenomenon the authors describe carefully: noise pushes data points near a curved boundary across it, and concave sections of that boundary collect more stray points than they lose, causing the effective boundary embedded in the data to drift away from its original position and gradually straighten. EDC, fitting the data rather than the hidden formula, tracks this shifted, smoothed boundary, achieving higher AUC than the very equation that generated the data. As noise grew from negligible to substantial across seventeen hundred artificial datasets, even a restricted linear version of EDC eventually beat the original nonlinear boundary.

The framework also proved capable of reconstructing notoriously difficult structures, including XOR-like and interaction-driven boundaries, and of fitting data produced by Gaussian clusters where no explicit target equation exists at all. On these cluster-based problems, which the authors describe as closest to real-world conditions among their artificial experiments, EDC outperformed existing symbolic classification approaches while landing near state-of-the-art black-box methods such as random forests, multi-layer perceptrons, and radial-basis-function support vector machines, all of which posted mean AUC values around 0.97.

Real-world benchmarks reinforced the pattern. Across nine binary classification datasets from the UCI repository, spanning tasks from banknote authentication to income prediction, EDC achieved a higher AUC than every competing equation-discovery-based classifier on every dataset, and beat simple decision trees across the board. A critical distance analysis showed that random forests, neural networks, and SVMs did not statistically significantly outperform EDC, even though those black-box methods won on several datasets, particularly ionosphere and sonar, where feature-class relationships appear to lie outside EDC’s current set of building blocks. On the Adult income dataset, the discovered equation offered a vivid demonstration of interpretability in action: one term acted as a penalty on years of education only when the individual appeared as a child in the household, while an exponential term over marital status effectively functioned as an if-else statement, adding roughly 28,586 to the score for married individuals living with a spouse and a negligible 8 otherwise.

The method’s main weakness is computational cost. In its default configuration, with a search depth of six and a beam width of ten, EDC takes dramatically longer than the sub-second runtimes of conventional classifiers, largely because pairwise interaction terms grow quadratically with the number of features and because one-hot encoding of categorical variables inflates the feature count. The authors show, however, that the expense is largely optional. Restricting the search depth, narrowing the beam, or replacing interaction terms with simple quadratic terms produced speed-ups of up to a factor of fifty at only a marginal loss of accuracy, with no statistically significant difference in performance across the benchmark suite. The authors also note that the framework is not limited to two classes: standard one-versus-rest schemes can extend EDC to multi-class problems, with each of the resulting equations explaining when a particular class label prevails.

The work arrives amid a broader movement toward explainable machine learning, driven by domains such as medicine, finance, and criminal justice where decisions carry real consequences and regulators increasingly demand justification. EDC offers a principled bridge between symbolic regression and classification, delivering models that are simultaneously compact, expressive, and auditable. While it may not dethrone random forests or deep networks on raw accuracy, it demonstrates that the gap is small enough, and the interpretability dividend large enough, that equations once again deserve a seat at the machine learning table. The authors have released their code, experimental setup, and datasets openly so that the entire study can be reproduced and the grammar extended by practitioners in new domains.

Subject of Research: Equation discovery for binary classification using interpretable symbolic decision boundaries

Article Title: Equation Discovery for Classification: Finding Interpretable Symbolic Specifications of the Decision Boundary

Article References: Toussaint, G., & Knobbe, A. (2026). Equation Discovery for Classification: Finding Interpretable Symbolic Specifications of the Decision Boundary. Machine Learning, 115(9), Article 215. https://doi.org/10.1007/s10994-026-07156-1

Image Credits: AI Generated

DOI: 10.1007/s10994-026-07156-1

Keywords: equation discovery, binary classification, symbolic regression, decision boundary, interpretable machine learning, beam search, symbolic classification, explainable AI, context-free grammar, UCI datasets, machine learning, Equation