Machine-learning models have demonstrated great success in learning complex patterns that enable them to make predictions about unobserved data. In addition to using models for prediction, the ability to interpret what a model has learned is receiving an increasing amount of attention. However, this increased focus has led to considerable confusion about the notion of interpretability. In particular, it is unclear how the wide array of proposed interpretation methods are related and what common concepts can be used to evaluate them. We aim to address these concerns by defining interpretability in the context of machine learning and introducing the predictive, descriptive, relevant (PDR) framework for discussing interpretations. The PDR framework provides 3 overarching desiderata for evaluation: predictive accuracy, descriptive accuracy, and relevancy, with relevancy judged relative to a human audience. Moreover, to help manage the deluge of interpretation methods, we introduce a categorization of existing techniques into model-based and post hoc categories, with subgroups including sparsity, modularity, and simulatability. To demonstrate how practitioners can use the PDR framework to evaluate and understand interpretations, we provide numerous real-world examples. These examples highlight the often underappreciated role played by human audiences in discussions of interpretability. Finally, based on our framework, we discuss limitations of existing methods and directions for future work. We hope that this work will provide a common vocabulary that will make it easier for both practitioners and researchers to discuss and choose from the full range of interpretation methods.interpretability | machine learning | explainability | relevancy
Anticancer drugs embedded in or conjugated with inert nanocarriers, referred to as nanomedicines, show many therapeutic advantages over free drugs, but the inert carrier materials are the major component (generally more than 90%) in nanomedicines, causing low drug loading contents and thus excessive uses of parenteral excipients. Herein, we demonstrate a new concept directly using drug molecules to fabricate nanocarriers in order to minimize use of inert materials, substantially increase the drug loading content, and suppress premature burst release. Taking advantage of the strong hydrophobicity of the anticancer drug camptothecin (CPT), one or two CPT molecule(s) were conjugated to a very short oligomer chain of ethylene glycol (OEG), forming amphiphilic phospholipid-mimicking prodrugs, OEG-CPT or OEG-DiCPT. The prodrugs formed stable liposome-like nanocapsules with a CPT loading content as high as 40 or 58 wt % with no burst release in aqueous solution. OEG-DiCPT released CPT once inside cells, which showed high in vitro and in vivo antitumor activity. Meanwhile, the resulting nanocapsules can be loaded with a water-soluble drug-doxorubicin salt (DOX.HCl)-with a high loading efficiency. The DOX.HCl-loaded nanocapsules simultaneously delivered two anticancer drugs, leading to a synergetic cytotoxicity to cancer cells. The concept directly using drugs as part of a carrier is applicable to fabricating other highly efficient nanocarriers with a substantially reduced use of inert carrier materials and increased drug loading content without premature burst release.
Reversing the charges: Targeted charge‐reversal nanoparticles (TCRNs) comprised of poly(ε‐caprolactone)‐block‐polyethyleneimine (PCL‐PEI), whose amine groups are converted into amides, are negatively charged at neutral pH but become positively charged at pH<6 (see picture). TCRNs effectively enter cells, regenerate the PEI layer in lysosomes, and localize in the nucleus for nuclear drug delivery.
DNA‐toxin anticancer drugs target nuclear DNA or its associated enzymes to elicit their pharmaceutical effects, but cancer cells have not only membrane‐associated but also many intracellular drug‐resistance mechanisms that limit their nuclear localization. Thus, delivering such drugs directly to the nucleus would bypass the drug‐resistance barriers. The cationic polymer poly(L‐lysine) (PLL) is capable of nuclear localization and may be used as a drug carrier for nuclear drug delivery, but its cationic charges make it toxic and cause problems in in‐vivo applications. Herein, PLL is used to demonstrate a pH‐triggered charge‐reversal carrier to solve this problem. PLL's primary amines are amidized as acid‐labile β‐carboxylic amides (PLL/amide). The negatively charged PLL/amide has a very low toxicity and low interaction with cells and, therefore, may be used in vivo. But once in cancer cells' acidic lysosomes, the acid‐labile amides hydrolyze into primary amines. The regenerated PLL escapes from the lysosomes and traverses into the nucleus. A cancer‐cell targeted nuclear‐localization polymer–drug conjugate has, thereby, been developed by introducing folic‐acid targeting groups and an anticancer drug camptothecin (CPT) to PLL/amide (FA‐PLL/amide‐CPT). The conjugate efficiently enters folate‐receptor overexpressing cancer cells and traverses to their nuclei. The CPT conjugated to the carrier by intracellular cleavable disulfide bonds shows much improved cytotoxicity.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.