The study, led by Mohammed Al-Ani of Hamad Bin Khalifa University in Qatar together with Siddhi P. Jani, Halima Bensmail, and Raghvendra Mall, builds on one of the most valuable datasets in leukemia research: the BeatAML cohort. That resource contains specimens from 805 patients with acute myeloid leukemia, each profiled for drug sensitivity across 165 different compounds in laboratory assays, alongside rich clinical annotations and both DNA and RNA sequencing data. Rather than treating this trove as a static catalog, the researchers turned it into training material for a model that learns how the molecular identity of a drug interacts with the cellular traits of a patient’s cancer.

What makes PREDICT-AML technically distinctive is the way it represents both sides of the drug-patient equation. On the drug side, the framework employs four complementary representations: classical physicochemical descriptors that capture properties such as molecular weight and lipophilicity; a novel set of knowledge-distilled embeddings, abbreviated KD-Embed, which compress what the model has learned about drug behavior into compact numerical vectors; and embeddings generated by two chemical language models, ChemBERTa and MolFormer, which are transformer architectures trained to read molecular structures the way language models read text. These representations allow the model to reason about drugs it has effectively seen in training data even when their chemical details differ substantially.

On the patient side, the model integrates several novel features that go beyond a simple list of mutations. The researchers engineered patient-specific oncogenic pathway proximity scores, which quantify how close a patient’s tumor expression profile sits to known cancer-driving signaling pathways. They also computed cell-state enrichment scores, which describe whether the leukemia cells resemble particular normal cell types, such as monocyte-like states, and they incorporated the patient’s genomic profile of somatic mutations. This combination of features lets the model capture the biological context in which a drug must act, rather than treating each patient as a bag of independent genetic alterations.

For the predictive engine itself, the team systematically searched for the best architecture using Optuna, a hyperparameter optimization framework, and identified TabPFN, a tabular foundation model, as the optimal choice. TabPFN represents a relatively new class of models designed specifically for the kind of structured, tabular data that dominates biomedical research, and its selection here underscores how foundation-model approaches are migrating from images and text into clinical prediction. When paired with the knowledge-distilled drug embeddings, TabPFN achieved a Pearson correlation of 0.690 between predicted and observed drug responses on a held-out test set, with a mean absolute error of 37.712.

One of the study’s most important methodological findings concerns how such models should be evaluated. The researchers found that random cross-validation, a common practice in machine learning studies, overestimates performance by approximately 13 percent relative to patient-stratified cross-validation, in which the model is tested on patients it has never encountered during training. Because the clinical goal is to predict responses for new patients rather than to memorize existing ones, the authors argue that patient-stratified evaluation is the appropriate benchmark. The distinction matters enormously for translational research: models that look impressive under random splits may fail when deployed on the next patient who walks through the clinic door.

The team did not stop at internal validation. They tested PREDICT-AML on two completely independent external datasets, LeeAML and FIMM-AML, generated at other institutions. The model achieved a Pearson correlation of 0.597 on LeeAML and an absolute correlation of 0.557 on FIMM-AML, confirming that its predictions generalize across patient populations, assay platforms, and clinical settings. In head-to-head comparisons, PREDICT-AML clearly outperformed established baselines: ElasticNet, a regularized regression method, managed only a correlation of 0.360, while MDREAM reached a Spearman correlation of 0.680. These results suggest that the combination of rich drug representations, multi-omics patient features, and the TabPFN architecture captures signal that simpler approaches miss.

Ablation experiments, in which the researchers systematically removed categories of input features, yielded biological insights of their own. Gene expression emerged as the single dominant predictor of drug response, reflecting the fact that a cell’s transcriptional state largely determines its vulnerability to any given compound. Perhaps more striking, the combination of somatic mutations with cell-state enrichment alone matched the performance of the full multi-omics feature set, hinting that a leaner panel of measurements could someday deliver comparable predictive power at lower cost. The knowledge-distilled embeddings proved to be the primary drug-identity signal, confirming the value of learning compact drug representations from large sensitivity datasets.

Interpretability analysis using SHAP values, a technique that quantifies each feature’s contribution to individual predictions, revealed recurring biological determinants of drug sensitivity across multiple compounds. Monocyte-like cell-state enrichment, expression of the BCL6 gene, and activity of the immunogenic cell death pathway all appeared repeatedly as markers of sensitivity. These findings are more than statistical curiosities; they point toward biological mechanisms that could be exploited therapeutically and suggest that the model is learning genuine biology rather than spurious correlations. For clinicians, such interpretability is essential before any predictive tool can inform treatment decisions.

The authors position PREDICT-AML as an interpretable, externally validated tool for individualized therapeutic prioritization in acute myeloid leukemia, a disease with limited therapeutic options and pressing clinical need. While the model’s predictions would still require prospective clinical validation before guiding real-world prescribing, the framework demonstrates that integrating chemical language models, tabular foundation models, and carefully engineered biological features can produce drug response predictions that hold up across institutions. As datasets like BeatAML continue to grow and similar resources accumulate for other cancers, the approach pioneered here may mark a step toward a future where a leukemia patient’s treatment is selected by an algorithm that has already seen, in molecular detail, how thousands of tumors responded to hundreds of drugs.

Subject of Research: Machine learning prediction of personalized drug response in acute myeloid leukemia using multi-omics data

Article Title: PREDICT-AML: personalized response via drug interaction with cellular traits for acute myeloid leukemia

Article References: Al-Ani, M., Jani, S. P., Bensmail, H., & Mall, R. (2026). PREDICT-AML: personalized response via drug interaction with cellular traits for acute myeloid leukemia. Journal of Translational Medicine. https://doi.org/10.1186/s12967-026-08817-4

Image Credits: AI Generated

DOI: 10.1186/s12967-026-08817-4

Keywords: acute myeloid leukemia, PREDICT-AML, machine learning, TabPFN, drug response prediction, multi-omics, BeatAML, chemical language models, precision oncology, SHAP interpretability, external validation, transcriptomics

The commercial stakes are considerable. The global fresh strawberry market has been estimated at roughly 20.7 billion US dollars in 2024, with projections exceeding 30 billion dollars by the mid-2030s as year-round demand grows. In the United Kingdom alone, official statistics value the strawberry crop at 389 million pounds for 2024. At daily resolution, even modest forecasting errors ripple through the entire operation: harvest crews are scheduled against predicted volumes, packing lines and storage capacity are booked in advance, and supermarkets hold contracts that assume reliable supply. A model that misses a peak picking day by a wide margin creates labour inefficiencies on one hand and post-harvest waste on the other.

What makes strawberries so hard to forecast is that daily yield is not a string of independent numbers but a biologically structured time series. Production follows a recognisable lifecycle: output starts at zero before harvest begins, climbs as plants enter peak fruiting, and falls back to zero as the season ends. That rise-peak-decline shape is governed by cultivar, planting density, accumulated temperature and humidity, and the decisions of farm managers. The researchers argue that a realistic forecasting framework must satisfy three conditions at once. It must preserve the natural lifecycle, including genuine zero-yield periods at season boundaries. It must enforce strict temporal alignment so that only information available before harvest can be used as a predictor. And it must test models on unseen future seasons rather than interpolating within the data it has already seen.

That last condition turns out to be the study’s sharpest methodological point. In agricultural machine learning, it is dangerously easy to inflate performance through information leakage, for example by letting weather recorded after a harvest event act as a predictor, or by using random train-test splits on strongly seasonal data. The team deliberately excluded aggregate productivity indicators such as yield per hectare and percentage picked from the feature set, because those variables indirectly encode the answer. Environmental sensor readings from the farm’s SoilMoistureSense platform, covering temperature and relative humidity, were aggregated strictly within calendar-day boundaries and merged by date so that no future information could slip into the pipeline.

The dataset itself is a portrait of messy commercial reality rather than a tidy laboratory experiment. It was assembled from three sources: annual weekly planning spreadsheets, 189 individual daily forecast files recording both the farm’s own operational predictions and actual harvested weights, and sub-daily environmental sensor exports. After cleaning and harmonisation across years, the modelling dataset contained 2,280 daily observations, which rose to 3,540 after the researchers appended seven zero-yield days at the start and end of each harvest cycle. That padding step, an ablation analysis showed, mattered a great deal: removing it degraded the best model’s mean absolute error from about 340 kilograms to nearly 611 kilograms, while extending padding to 14 days lowered absolute error further but reduced explained variance. The seven-day configuration offered the best balance.

Feature engineering was deliberately biological. Rolling averages of prior yield over 1, 3, 7, 14 and 21 days captured what the authors call production memory, the tendency of strawberry harvests to persist across neighbouring days as overlapping cohorts of fruit mature and ripen. Rolling seven-day temperature averages and 14-day humidity averages represented accumulated environmental exposure, reflecting how heat and moisture stress modulate fruit development. Plant counts from the planning spreadsheets served as a static capacity indicator for each field. Together these features let the models learn whether production is rising, plateauing or declining, without ever violating temporal causality.

The model comparison spanned thirteen approaches, from simple ridge regression through gradient boosting frameworks such as XGBoost, LightGBM and CatBoost, to compact neural networks including a multilayer perceptron, a small LSTM and a temporal convolutional network, and finally TabPFN, a pretrained tabular foundation model. The results were striking. Under a conventional random 80/20 split, TabPFN achieved the lowest error with a mean absolute error of 340.18 kilograms and an R-squared of 0.733, while the neural sequence models performed worse than a constant mean predictor, producing negative R-squared values. Leave-one-out cross-validation told the same story, with TabPFN again leading at a mean absolute error of 338.09 kilograms. The authors note that with only six seasons of data, the strong inductive biases of tree ensembles and pretrained tabular models matter far more than architectural depth, which is precisely why the from-scratch deep learning baselines struggled.

The most deployment-relevant test, however, was forward-year validation: train on 2020 through 2024, then predict the entirely unseen 2025 season. Under this protocol TabPFN remained the strongest performer with a mean absolute error of 438.53 kilograms and an R-squared of 0.696, followed by Random Forest, Histogram Gradient Boosting and Extra Trees. The rise in error relative to the random split is not a failure, the researchers argue, but an honest measure of genuine interannual variation in weather, phenology and management. Permutation importance analysis on the 2025 test season confirmed that the models rely on biologically sensible variables: recent yield lags dominate the ranking, and interaction features combining recent production with temperature and humidity contribute measurably, suggesting that weather modulates an established production trajectory rather than acting as an independent driver.

The study is candid about its limits. All data come from a single farm, so the work demonstrates temporal generalisation to an unseen season within one commercial setting, not transfer across farms or regions. The available sensors captured only temperature and humidity, and the authors identify the absence of solar radiation, day length, irrigation records and growing degree days as a likely explanation for a recurring error mode: the systematic underestimation of peak harvest days, when similar recent conditions can conceal very different yield potential. Adding phenology-aware signals such as flower or truss counts, whether recorded manually or by computer vision, is flagged as a plausible route to better peak capture.

Even so, the implications are significant for an industry built on thin margins and perishable goods. The findings suggest that daily yield forecasting can be integrated into commercial decision-support systems using the data farms already collect, without exotic sensors or vast historical archives. In small, discontinuous seasonal datasets, the study concludes, strong inductive bias and regularisation beat model complexity, and rigorous forward-year evaluation is the only honest yardstick of what a forecasting system will actually deliver when the next season begins. For growers weighing the promise of agricultural AI, that combination of realism and rigour may be the most valuable harvest of all.

Subject of Research: Daily strawberry yield forecasting with artificial intelligence in commercial horticulture

Article Title: Daily strawberry yield forecasting using artificial intelligence in commercial horticulture

Article References: Ahmed, S., & Caldwell, N. H. (2026). Daily strawberry yield forecasting using artificial intelligence in commercial horticulture. Smart Agricultural Technology, 15, Article 102583. https://doi.org/10.1016/j.atech.2026.102583

Image Credits: AI Generated

DOI: 10.1016/j.atech.2026.102583

Keywords: strawberry, yield forecasting, machine learning, TabPFN, deep learning, precision agriculture, time series, commercial horticulture, feature engineering, forward-year validation, IoT sensors, UK farming

The study, published in Environmental and Sustainability Indicators, introduces a framework that fuses machine learning with satellite-based environmental monitoring to construct a dynamic baseline of household vulnerability and capability along three distinct stretches of the highway. Its central insight is deceptively simple: because road projects of this scale are built in segmented phases for administrative and logistical convenience, pre-development, construction and post-development conditions exist simultaneously on different sections of the same corridor. Traditional assessment designs, which assume a sequential project lifecycle, collapse under this temporal parallelism, fragmenting impact information precisely where it is most needed. The author calls this methodological stagnation a pathway to development paralysis, a state in which planners cannot understand current habitat-level interactions and therefore cannot anticipate stakeholder grievances or design effective policy.

To break that paralysis, the research surveyed 462 households across three field clusters along the highway axis, each representing a different phase of development: a pre-development stretch, an active construction zone and a post-development segment. Using a multi-stage purposive sampling design grounded in standard Indian environmental impact assessment buffer guidelines, the study grouped villages sharing common valley habitats into development clusters and conducted a census-style digital survey using Open Data Kit. Forty-nine discrete indicators of household wellbeing were processed into a cumulative capability index, drawing on Anderson and Woodrow’s Capacities and Vulnerabilities framework and the Alkire-Foster multidimensional methodology, and spanning six themes: access to social welfare, development goods, economic stability, human capital, public health access, and ownership of land and immovable property.

The analytical heart of the study lies in its use of interpretable tabular machine learning models, TabNet and TabPFN, to identify which indicators truly drive differences in household capability. TabNet employs sequential attention masks, including Sparsemax and Entmax, to isolate significant features from high-dimensional socio-economic data, while TabPFN, a pretrained tabular foundation model, handles small datasets with remarkable efficiency. When benchmarked against traditional classifiers, the results were striking. Random Forest managed only 52 per cent accuracy, XGBoost 49 per cent, and SVM 69 per cent, but TabPFN achieved 86 per cent accuracy and an identical macro F1 score, with a macro ROC-AUC of 0.986. Crucially, its misclassifications were almost entirely ordinal, leaking only between adjacent capability classes, which strengthens confidence that the model captures a genuine underlying gradient of household resilience rather than noise.

The capability scores themselves told a nuanced story. Among the 462 households surveyed, composite scores ranged from 13 to 39 out of 49 possible indicators, following a roughly normal distribution in which more than 71 per cent of households possessed at least half of the measured capabilities. Cross-thematic analysis revealed that social assets and economic stability were moderately correlated, suggesting that economic status manifests largely through asset ownership, while education access stood apart with no significant linear relationship to other themes. At the indicator level, access to television media, reliable electricity supply for more than 12 hours a day, emergency healthcare access and female job-preference education emerged as the most powerful discriminants of relative capability across the region, pointing to the central importance of information access, energy reliability and gendered educational attainment.

Perhaps the most compelling finding is that the drivers of household capability change systematically across the project lifecycle, a pattern pooled analyses would completely obscure. In the pre-development cluster, capability differences were concentrated in pre-existing endowments, with television ownership and emergency healthcare access jointly accounting for roughly half of the cumulative top-ten feature weights. During active construction, the importance distribution flattened dramatically, with ten variables clustered within a narrow range and out-of-pocket health expenditure emerging as the leading predictor, alongside housing-quality indicators such as roof material and settlement-stability measures including residency permanency and potable water access. This diffuse structure is consistent with communities absorbing construction-phase disruptions through many small, exposure-linked stresses rather than a few dominant shocks. In the post-development cluster, house ownership, absent from both earlier phases, became the single largest predictor, a pattern the author links to redevelopment of immovable assets and capacity addition in hospitality infrastructure such as homestays along the completed highway.

Only two indicators persisted across all three phases: out-of-pocket health expenditure and female education for employment. Their consistent salience suggests that health-cost burdens and gendered educational attainment function as structural constraints on capability that are largely orthogonal to the project cycle itself. The study reinforces this with perturbation testing: when Gaussian noise of 5 per cent variance was applied to the self-reported household data, 17 per cent of households reclassified, indicating that a large share of the population sits precariously on the boundary of its current capability class. At 10 per cent variance, nearly a quarter of households shifted, and the direction of movement reversed, with more households moving up than down, a threshold effect suggesting that minor shocks degrade household standing while larger perturbations may trigger compensatory mechanisms within communities.

Parallel to the machine learning analysis, the study tracked environmental change using Sentinel-2 imagery at 10-metre resolution across four temporal benchmarks from 2016 to 2025, processed through Google Earth Engine. Land-use and land-cover analysis revealed a persistent, cluster-independent trade-off between forest and agriculture, with forest cover inversely correlated with agricultural extent in every spatial subset, reaching a global correlation of negative 0.90. Built-up area sustained positive growth across the full period, expanding by 2.03 per cent, with the sharpest increase of 1.72 percentage points occurring between 2019 and 2022, potentially reflecting capacity addition in tourism and hospitality assets during the pandemic lockdown phases. In the construction cluster, built-up expansion correlated strongly with agricultural contraction and water-body decline, signs that excavation and muck-dumping works are disturbing spring paths that sustain agriculture-based livelihoods in this landslide-prone landscape.

The convergence of independently derived social and spatial evidence is what gives the framework its diagnostic power. In the construction cluster, where machine learning flagged flat capability distributions dominated by exposure-linked indicators such as private health affordability and potable water access, satellite data simultaneously recorded the strongest statistical associations between infrastructure growth and environmental degradation. In the post-development cluster, where house ownership dominated capability predictions, the spatial analysis documented transitions in which agricultural land passed through an intermediate degraded, fallow stage before conversion to built-up use, rather than being directly substituted. High-capability households in the post-development zone clustered neatly around the highway axis, while high-capability zones in the earlier phases appeared as inherited strengths rather than project-derived gains, a distinction with direct implications for how planners interpret baseline data.

The study’s implications reach well beyond Uttarakhand. It demonstrates that household resilience is not time-invariant, that capability drivers migrate from asset endowments through exposure-linked drains during construction before consolidating into spatialised real-estate assets, and that community systems exhibit threshold effects under perturbation. For a sector in which development imperatives routinely outpace assessment mechanisms, and in which template-driven, context-agnostic impact reports have been repeatedly criticised for externalising local vulnerabilities, the approach offers a way to back-propagate learnings from completed segments into stretches where construction has not yet begun. The author cautions that machine learning on small datasets carries overfitting risks, and that data-supported policy decisions still require stakeholder engagement and in-situ validation. Yet the framework’s promise is clear: a dynamic, socially embedded diagnostic capable of identifying deficit clusters and guiding micro-level interventions as the socio-ecological realities of affected communities evolve along one of the world’s most consequential mountain corridors.

Subject of Research: Socio-environmental impact assessment of the Char Dham expressway in Uttarakhand, India, using machine learning and satellite land-cover analysis.

Article Title: The baseline of fragility and capability: Hedging socio-environmental impacts of Char Dham expressway in Uttarakhand, India

Article References: Kumar, S. (2026). The baseline of fragility and capability: Hedging socio-environmental impacts of Char Dham expressway in Uttarakhand, India. Environmental and Sustainability Indicators, 32, Article 101510. https://doi.org/10.1016/j.indic.2026.101510

Image Credits: AI Generated

DOI: 10.1016/j.indic.2026.101510

Keywords: Char Dham Expressway, Uttarakhand, impact assessment, machine learning, TabPFN, capability index, land use change, Himalayan fragility, household resilience, Sentinel-2, sustainability indicators, temporal parallelism