{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,30]],"date-time":"2026-07-30T17:48:25Z","timestamp":1785433705737,"version":"3.56.0"},"reference-count":60,"publisher":"MDPI AG","issue":"8","license":[{"start":{"date-parts":[[2026,7,24]],"date-time":"2026-07-24T00:00:00Z","timestamp":1784851200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/100011495","name":"Massachusetts Life Sciences Center","doi-asserted-by":"crossref","id":[{"id":"10.13039\/100011495","id-type":"DOI","asserted-by":"crossref"}]},{"id":[{"id":"https:\/\/ror.org\/033km2469","id-type":"ROR","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000002","name":"National Institutes of Health","doi-asserted-by":"crossref","award":["R01HL150394-SI"],"award-info":[{"award-number":["R01HL150394-SI"]}],"id":[{"id":"10.13039\/100000002","id-type":"DOI","asserted-by":"crossref"}]},{"DOI":"10.13039\/100000002","name":"National Institutes of Health","doi-asserted-by":"crossref","award":["R01AI148747"],"award-info":[{"award-number":["R01AI148747"]}],"id":[{"id":"10.13039\/100000002","id-type":"DOI","asserted-by":"crossref"}]},{"DOI":"10.13039\/100000002","name":"National Institutes of Health","doi-asserted-by":"crossref","award":["R01AI148747-SI"],"award-info":[{"award-number":["R01AI148747-SI"]}],"id":[{"id":"10.13039\/100000002","id-type":"DOI","asserted-by":"crossref"}]},{"award":["R01HL150394-SI"],"award-info":[{"award-number":["R01HL150394-SI"]}],"id":[{"id":"https:\/\/ror.org\/045p44t13","id-type":"ROR","asserted-by":"publisher"}]},{"award":["R01AI148747"],"award-info":[{"award-number":["R01AI148747"]}],"id":[{"id":"https:\/\/ror.org\/045p44t13","id-type":"ROR","asserted-by":"publisher"}]},{"award":["R01AI148747-SI"],"award-info":[{"award-number":["R01AI148747-SI"]}],"id":[{"id":"https:\/\/ror.org\/045p44t13","id-type":"ROR","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000936","name":"Gordon and Betty Moore Foundation","doi-asserted-by":"crossref","id":[{"id":"10.13039\/100000936","id-type":"DOI","asserted-by":"crossref"}]},{"id":[{"id":"https:\/\/ror.org\/006wxqw41","id-type":"ROR","asserted-by":"publisher"}]},{"name":"Tianqiao and Chrissy Chen Institute"},{"name":"Food and Drug Administration"},{"id":[{"id":"https:\/\/ror.org\/05hvrsg85","id-type":"ROR","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Bioengineering"],"abstract":"<jats:p>In deep-learning-based image classification, achieving high performance requires diverse training sets. However, the current best practice\u2014maximizing dataset size and class balance\u2014does not guarantee dataset diversity. We hypothesized that, for a given model architecture, performance improves by maximizing diversity more directly. To test this hypothesis, we introduce a comprehensive framework of \u201csentropic\u201d diversity measures or \u201csentropies\u201d from ecology that generalizes familiar quantities like Shannon entropy by accounting for similarities among images. Size and class balance emerge as special cases. Analyzing thousands of subsets from seven medical-imaging datasets showed that the best correlates of performance were not size or class balance but A\u2014\u201cbig alpha\u201d\u2014a set of sentropies interpreted as the effective number of image\u2013class pairs in the dataset, after accounting for similarities among images. One of these, A0, explained 67% of the variance in balanced accuracy, vs. 54% for class balance and just 39% for size. The best pair of measures was size-plus-A1 (79%), which outperformed size-plus-class-balance (74%). Subsets with the largest A0 performed up to 16% better than those with the largest size (median improvement, 8%). We propose developing methods to maximize A as a way to improve deep learning performance in medical imaging.<\/jats:p>","DOI":"10.3390\/bioengineering13080854","type":"journal-article","created":{"date-parts":[[2026,7,24]],"date-time":"2026-07-24T07:36:13Z","timestamp":1784878573000},"page":"854","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Beyond Size and Class Balance: Alpha as a New Dataset Quality Metric for Deep Learning"],"prefix":"10.3390","volume":"13","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-7416-5858","authenticated-orcid":false,"given":"Josiah D.","family":"Couch","sequence":"first","affiliation":[{"name":"Department of Pathology, Beth Israel Deaconess Medical Center (BIDMC), Boston, MA 02215, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7134-0040","authenticated-orcid":false,"given":"Rima","family":"Arnaout","sequence":"additional","affiliation":[{"name":"Department of Medicine, University of California San Francisco, San Francisco, CA 94143, USA"},{"name":"Bakar Institute for Computational Health Sciences, University of California San Francisco, San Francisco, CA 94143, USA"},{"name":"Center for Intelligent Imaging, University of California San Francisco, San Francisco, CA 94143, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6955-9310","authenticated-orcid":false,"given":"Ramy","family":"Arnaout","sequence":"additional","affiliation":[{"name":"Department of Pathology, Beth Israel Deaconess Medical Center (BIDMC), Boston, MA 02215, USA"},{"name":"Division of Clinical Informatics, Department of Medicine, Beth Israel Deaconess Medical Center (BIDMC), Boston, MA 02215, USA"},{"name":"Harvard Medical School, Harvard University, Boston, MA 02115, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2026,7,24]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"1061","DOI":"10.1161\/CIR.0000000000001173","article-title":"Principles for Health Information Collection, Sharing, and Use: A Policy Statement from the American Heart Association","volume":"148","author":"Armoundas","year":"2023","journal-title":"Circulation"},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"63","DOI":"10.1016\/j.jacc.2023.10.025","article-title":"Novel Techniques in Imaging Congenital Heart Disease: JACC Scientific Statement","volume":"83","author":"Sachdeva","year":"2024","journal-title":"J. Am. Coll. Cardiol."},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"1209","DOI":"10.1016\/j.jcmg.2023.05.012","article-title":"Proceedings of the NHLBI Workshop on Artificial Intelligence in Cardiovascular Imaging: Translation to Patient Care","volume":"16","author":"Dey","year":"2023","journal-title":"JACC Cardiovasc. Imaging"},{"key":"ref_4","unstructured":"Simard, P., Victorri, B., LeCun, Y., and Denker, J. (1991). Tangent Prop\u2014A formalism for specifying selected invariances in an adaptive network. Adv. Neural Inf. Process. Syst., 4, Available online: https:\/\/proceedings.neurips.cc\/paper\/1991\/hash\/65658fde58ab3c2b6e5132a39fae7cb9-Abstract.html."},{"key":"ref_5","unstructured":"Mozer, M., Jordan, M., and Petsche, T. (1996). Effective Training of a Neural Network Character Classifier for Word Recognition. Proceedings of the Advances in Neural Information Processing Systems, MIT Press."},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"60","DOI":"10.1186\/s40537-019-0197-0","article-title":"A survey on Image Data Augmentation for Deep Learning","volume":"6","author":"Shorten","year":"2019","journal-title":"J. Big Data"},{"key":"ref_7","unstructured":"Oh, J., Kwon, C., and Cho, S. (1995). Learning algorithms for classification: A comparison on handwritten digit recognition. Proceedings of the Neural Networks, World Scientific."},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"249","DOI":"10.1016\/j.neunet.2018.07.011","article-title":"A systematic study of the class imbalance problem in convolutional neural networks","volume":"106","author":"Buda","year":"2018","journal-title":"Neural Netw."},{"key":"ref_9","doi-asserted-by":"crossref","unstructured":"Athalye, C., and Arnaout, R. (2023). Domain-guided data augmentation for deep learning on medical imaging. PLoS ONE, 18.","DOI":"10.1371\/journal.pone.0282532"},{"key":"ref_10","unstructured":"Renyi, A. (July, January 20). On measures of entropy and information. Proceedings of the Fourth Berkeley Symposium on Mathematical Statistics and Probability, Berkeley, CA, USA."},{"key":"ref_11","doi-asserted-by":"crossref","first-page":"379","DOI":"10.1002\/j.1538-7305.1948.tb01338.x","article-title":"A mathematical theory of communication","volume":"27","author":"Shannon","year":"1948","journal-title":"Bell Syst. Tech. J."},{"key":"ref_12","doi-asserted-by":"crossref","first-page":"1270","DOI":"10.1038\/s41592-021-01302-4","article-title":"The class imbalance problem","volume":"18","author":"Megahed","year":"2021","journal-title":"Nat. Methods"},{"key":"ref_13","first-page":"111","article-title":"The class imbalance problem: Significance and strategies","volume":"56","author":"Japkowicz","year":"2000","journal-title":"Proc. Int. Conf. Artif. Intell."},{"key":"ref_14","first-page":"813","article-title":"Concept Learning and the Problem of Small Disjuncts","volume":"89","author":"Holte","year":"1989","journal-title":"Int. Jt. Conf. Artif. Intell."},{"key":"ref_15","first-page":"67","article-title":"Concept-Learning in the Presence of Between-Class and Within-Class Imbalances","volume":"2056","author":"Japkowicz","year":"2001","journal-title":"Adv. Artif. Intell."},{"key":"ref_16","doi-asserted-by":"crossref","first-page":"40","DOI":"10.1145\/1007730.1007737","article-title":"Class imbalances versus small disjuncts","volume":"6","author":"Jo","year":"2004","journal-title":"ACM SIGKDD Explor. Newsl."},{"key":"ref_17","doi-asserted-by":"crossref","unstructured":"Ojima, Y., Horiuchi, S., and Ishikawa, F. (2021, January 23\u201326). Model-based Data-Complexity Estimator for Deep Learning Systems. Proceedings of the 2021 IEEE International Conference on Artificial Intelligence Testing (AITest), Oxford, UK.","DOI":"10.1109\/AITEST52744.2021.00011"},{"key":"ref_18","doi-asserted-by":"crossref","unstructured":"Cho, H., and Lee, S. (2021). Data Quality Measures and Efficient Evaluation Algorithms for Large-Scale High-Dimensional Data. Appl. Sci., 11.","DOI":"10.3390\/app11020472"},{"key":"ref_19","doi-asserted-by":"crossref","unstructured":"Park, L.A.F., Gomes, H.M., Doborjeh, M., Boo, Y.L., Koh, Y.S., Zhao, Y., Williams, G., and Simoff, S. (2022). Measuring Difficulty of Learning Using Ensemble Methods. Proceedings of the Australasian Conference on Data Mining, Parramatta, Australia, 12\u201315 December 2022, Springer Nature.","DOI":"10.1007\/978-981-19-8746-5"},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"3085","DOI":"10.1007\/s10994-022-06205-9","article-title":"Relating instance hardness to classification performance in a dataset: A visual approach","volume":"111","author":"Paiva","year":"2022","journal-title":"Mach. Learn."},{"key":"ref_21","doi-asserted-by":"crossref","unstructured":"Ivanovici, M., Coliban, R.M., Hatfaludi, C., and Nicolae, I.E. (2020). Color Image Complexity versus Over-Segmentation: A Preliminary Study on the Correlation between Complexity Measures and Number of Segments. J. Imaging, 6.","DOI":"10.3390\/jimaging6040016"},{"key":"ref_22","unstructured":"Wan, Z., Wang, Z., Chung, C., and Wang, Z. (2023). A Survey of Data Optimization for Problems in Computer Vision Datasets. arXiv."},{"key":"ref_23","doi-asserted-by":"crossref","unstructured":"He, Y., Xiao, L., and Zhou, J.T. (2023). You Only Condense Once: Two Rules for Pruning Condensed Datasets. arXiv.","DOI":"10.52202\/075280-1710"},{"key":"ref_24","unstructured":"Chen, M., Huang, B., Lu, J., Li, B., Wang, Y., Cheng, M., and Wang, W. (2023). Dataset Distillation via Adversarial Prediction Matching. arXiv."},{"key":"ref_25","doi-asserted-by":"crossref","unstructured":"Sundar, A., Keskin, G., Chandak, C., Chen, I.F., Ghahremani, P., and Ghosh, S. (2023, January 4\u201310). Prune then distill: Dataset distillation with importance sampling. Proceedings of the ICASSP 2023\u20142023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Rhodes Island, Greece.","DOI":"10.1109\/ICASSP49357.2023.10095172"},{"key":"ref_26","doi-asserted-by":"crossref","first-page":"30","DOI":"10.1145\/1007730.1007736","article-title":"Learning from imbalanced data sets with boosting and data generation: The DataBoost-IM approach","volume":"6","author":"Guo","year":"2004","journal-title":"SIGKDD Explor. Newsl."},{"key":"ref_27","doi-asserted-by":"crossref","first-page":"1079","DOI":"10.1093\/jamia\/ocad055","article-title":"ENRICHing medical imaging training sets enables more efficient machine learning","volume":"30","author":"Chinn","year":"2023","journal-title":"J. Am. Med. Inform. Assoc."},{"key":"ref_28","doi-asserted-by":"crossref","unstructured":"Aggarwal, U., Popescu, A., and Hudelot, C. (2020). Active Learning for Imbalanced Datasets. Proceedings of the 2020 IEEE Winter Conference on Applications of Computer Vision (WACV), IEEE.","DOI":"10.1109\/WACV45572.2020.9093475"},{"key":"ref_29","unstructured":"Zhan, X., Wang, Q., Huang, K.H., Xiong, H., Dou, D., and Chan, A.B. (2022). A Comparative Survey of Deep Active Learning. arXiv."},{"key":"ref_30","unstructured":"Touretzky, D. (1988). Scaling and Generalization in Neural Networks: A Case Study. Advances in Neural Information Processing Systems, Morgan-Kaufmann."},{"key":"ref_31","unstructured":"Hestness, J., Narang, S., Ardalani, N., Diamos, G., Jun, H., Kianinejad, H., Patwary, M.M.A., Yang, Y., and Zhou, Y. (2017). Deep Learning Scaling is Predictable, Empirically. arXiv."},{"key":"ref_32","unstructured":"Kaplan, J., McCandlish, S., Henighan, T., Brown, T.B., Chess, B., Child, R., Gray, S., Radford, A., Wu, J., and Amodei, D. (2020). Scaling Laws for Neural Language Models. arXiv."},{"key":"ref_33","doi-asserted-by":"crossref","unstructured":"Althnian, A., AlSaeed, D., Al-Baity, H., Samha, A., Dris, A.B., Alzakari, N., Abou Elwafa, A., and Kurdi, H. (2021). Impact of Dataset Size on Classification Performance: An Empirical Evaluation in the Medical Domain. Appl. Sci., 11.","DOI":"10.3390\/app11020796"},{"key":"ref_34","doi-asserted-by":"crossref","first-page":"6056","DOI":"10.1103\/PhysRevA.45.6056","article-title":"Statistical mechanics of learning from examples","volume":"45","author":"Seung","year":"1992","journal-title":"Phys. Rev. A"},{"key":"ref_35","doi-asserted-by":"crossref","first-page":"e2311878121","DOI":"10.1073\/pnas.2311878121","article-title":"Explaining Neural Scaling Laws","volume":"121","author":"Bahri","year":"2024","journal-title":"Proc. Natl. Acad. Sci. USA"},{"key":"ref_36","unstructured":"Maloney, A., Roberts, D.A., and Sully, J. (2022). A Solvable Model of Neural Scaling Laws. arXiv."},{"key":"ref_37","doi-asserted-by":"crossref","unstructured":"Zhang, Z. (2024). Neural Scaling Laws From Large-N Field Theory: Solvable Model Beyond the Ridgeless Limit. arXiv.","DOI":"10.1088\/2632-2153\/adc872"},{"key":"ref_38","doi-asserted-by":"crossref","unstructured":"Sorscher, B., Geirhos, R., Shekhar, S., Ganguli, S., and Morcos, A.S. (2022). Beyond neural scaling laws: Beating power law scaling via data pruning. arXiv.","DOI":"10.52202\/068431-1419"},{"key":"ref_39","unstructured":"Dohmatob, E., Pezeshki, M., and Askari-Hemmat, R. (2025). Why Less is More (Sometimes): A Theory of Data Curation. arXiv."},{"key":"ref_40","doi-asserted-by":"crossref","first-page":"477","DOI":"10.1890\/10-2402.1","article-title":"Measuring diversity: The importance of species similarity","volume":"93","author":"Leinster","year":"2012","journal-title":"Ecology"},{"key":"ref_41","unstructured":"Reeve, R., Leinster, T., Cobbold, C.A., Thompson, J., Brummitt, N., Mitchell, S.N., and Matthews, L. (2016). How to partition diversity. arXiv."},{"key":"ref_42","doi-asserted-by":"crossref","unstructured":"Leinster, T. (2022). Entropy and Diversity: The Axiomatic Approach. arXiv.","DOI":"10.1017\/9781108963558"},{"key":"ref_43","unstructured":"Friedman, D., and Bousso Dieng, A. (2022). The Vendi Score: A Diversity Evaluation Metric for Machine Learning. arXiv."},{"key":"ref_44","unstructured":"Pasarkar, A.P., and Dieng, A.B. (2025). The Vendiscope: An Algorithmic Microscope For Data Collections. arXiv."},{"key":"ref_45","unstructured":"Pasarkar, A.P., and Dieng, A.B. (2023). Cousins Of The Vendi Score: A Family Of Similarity-Based Diversity Metrics For Science And Machine Learning. arXiv."},{"key":"ref_46","doi-asserted-by":"crossref","first-page":"427","DOI":"10.2307\/1934352","article-title":"Diversity and Evenness: A Unifying Notation and Its Consequences","volume":"54","author":"Hill","year":"1973","journal-title":"Ecology"},{"key":"ref_47","unstructured":"LeCun, Y., and Cortes, C. (2026, July 21). The MNIST Database of Handwritten Digits. Available online: http:\/\/yann.lecun.com\/exdb\/mnist\/."},{"key":"ref_48","doi-asserted-by":"crossref","first-page":"2427","DOI":"10.1890\/06-1736.1","article-title":"Partitioning diversity into independent alpha and beta components","volume":"88","author":"Jost","year":"2007","journal-title":"Ecology"},{"key":"ref_49","unstructured":"Nguyen, P., Arora, R., Hill, E.D., Braun, J., Morgan, A., Quintana, L.M., Mazzoni, G., Lee, G.R., Arnaout, R., and Arnaout, R. (2023). Sentropy: A Python Package for Revealing Hidden Differences in Complex Datasets. arXiv."},{"key":"ref_50","doi-asserted-by":"crossref","first-page":"055305","DOI":"10.1103\/n2s3-2g85","article-title":"Comparing frameworks for similarity-aware generalized entropy (sentropy)","volume":"113","author":"Nguyen","year":"2026","journal-title":"Phys. Rev. E"},{"key":"ref_51","doi-asserted-by":"crossref","unstructured":"Yang, J., Shi, R., and Ni, B. (2021, January 13\u201316). MedMNIST Classification Decathlon: A Lightweight AutoML Benchmark for Medical Image Analysis. Proceedings of the IEEE 18th International Symposium on Biomedical Imaging (ISBI), Nice, France.","DOI":"10.1109\/ISBI48211.2021.9434062"},{"key":"ref_52","doi-asserted-by":"crossref","first-page":"41","DOI":"10.1038\/s41597-022-01721-8","article-title":"MedMNIST v2-A large-scale lightweight benchmark for 2D and 3D biomedical image classification","volume":"10","author":"Yang","year":"2023","journal-title":"Sci. Data"},{"key":"ref_53","doi-asserted-by":"crossref","first-page":"102680","DOI":"10.1016\/j.media.2022.102680","article-title":"The Liver Tumor Segmentation Benchmark (LiTS)","volume":"84","author":"Bilic","year":"2023","journal-title":"Med. Image Anal."},{"key":"ref_54","doi-asserted-by":"crossref","first-page":"882","DOI":"10.1038\/s41591-021-01342-5","article-title":"An ensemble of neural networks provides expert-level prenatal detection of complex congenital heart disease","volume":"27","author":"Arnaout","year":"2021","journal-title":"Nat. Med."},{"key":"ref_55","doi-asserted-by":"crossref","first-page":"6","DOI":"10.1038\/s41746-017-0013-1","article-title":"Fast and accurate view classification of echocardiograms using deep learning","volume":"1","author":"Madani","year":"2018","journal-title":"npj Digit. Med."},{"key":"ref_56","doi-asserted-by":"crossref","first-page":"132665","DOI":"10.1109\/ACCESS.2020.3010287","article-title":"Can AI help in screening Viral and COVID-19 pneumonia?","volume":"8","author":"Chowdhury","year":"2020","journal-title":"IEEE Access"},{"key":"ref_57","doi-asserted-by":"crossref","first-page":"104319","DOI":"10.1016\/j.compbiomed.2021.104319","article-title":"Exploring the effect of image enhancement techniques on COVID-19 detection using chest X-ray images","volume":"132","author":"Rahman","year":"2021","journal-title":"Comput. Biol. Med."},{"key":"ref_58","doi-asserted-by":"crossref","unstructured":"He, K., Zhang, X., Ren, S., and Sun, J. (2015). Deep Residual Learning for Image Recognition. arXiv.","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref_59","unstructured":"Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., and Antiga, L. (2019). PyTorch: An Imperative Style, High-Performance Deep Learning Library. arXiv."},{"key":"ref_60","first-page":"2825","article-title":"Scikit-learn: Machine Learning in Python","volume":"12","author":"Pedregosa","year":"2011","journal-title":"J. Mach. Learn. Res."}],"container-title":["Bioengineering"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2306-5354\/13\/8\/854\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,26]],"date-time":"2026-07-26T04:12:58Z","timestamp":1785039178000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2306-5354\/13\/8\/854"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,24]]},"references-count":60,"journal-issue":{"issue":"8","published-online":{"date-parts":[[2026,8]]}},"alternative-id":["bioengineering13080854"],"URL":"https:\/\/doi.org\/10.3390\/bioengineering13080854","relation":{},"ISSN":["2306-5354"],"issn-type":[{"value":"2306-5354","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,24]]}}}