The Value of Scientific Publications in Surgical Device Innovation.

BACKGROUND: Clinical trials represent a significant risk in the commercialization of surgical technologies. There is incentive for companies to mitigate their regulatory risk by targeting 510K over Premarket Approval (PMA) pathways in order to limit the scope, complexity and cost of clinical trials. As such, not all companies will publish clinical data in the scientific literature. PURPOSE: We set out to investigate the relationship between scientific publication by surgical device companies and the impact it has on company valuation. We hypothesize that publishing in the scientific literature correlates with success of the surgical device companies as measured by funding. RESEARCH DESIGN: We first obtained a list of surgical device startup companies and their financial deals using the Pitchbook database. Those companies were then cross referenced with the FDA database and the Dimensions database for product registrations and peer reviewed publications, respectively. Analysis was then performed using these query results. STUDY SAMPLE AND DATA COLLECTION: We obtained a list of US surgical device startups financing deals closed between 2010 and 2020 from the Pitchbook database. We queried the Pitchbook for deal dates from January 1, 2010 to January 1, 2020 for deal types spanning early stage investment to IPO. Deals were limited to those conducted in the United States and to the surgical device industry. We queried the FDA database for product registration information associated with each of the companies involved in the deals. We tabulated the number of journal articles associated with surgical device companies using the Dimensions Search API as well as a manual confirmation. RESULTS: Five hundred thirty five (535) deals from 222 companies were found in Pitchbook that met our criteria. Querying the FDA database resulted in 578 registrations associated with these companies. Publications per company ranged widely. CONCLUSIONS: Companies that are able to generate a more numerous publications had correspondingly higher valuations during funding rounds. A subset of outstanding companies were analyzed and at least four factors affect: direct value of publications, indirect valve of publications, survivorship bias, and adoption share; each of which will be discussed in this manuscript.

Shallow Representation Learning via Kernel PCA Improves QSAR Modelability.

Linear models offer a robust, flexible, and computationally efficient set of tools for modeling quantitative structure-activity relationships (QSARs) but have been eclipsed in performance by nonlinear methods. Support vector machines (SVMs) and neural networks are currently among the most popular and accurate QSAR methods because they learn new representations of the data that greatly improve modelability. In this work, we use shallow representation learning to improve the accuracy of L1 regularized logistic regression (LASSO) and meet the performance of Tanimoto SVM. We embedded chemical fingerprints in Euclidean space using Tanimoto (a.k.a. Jaccard) similarity kernel principal component analysis (KPCA) and compared the effects on LASSO and SVM model performance for predicting the binding activities of chemical compounds against 102 virtual screening targets. We observed similar performance and patterns of improvement for LASSO and SVM. We also empirically measured model training and cross-validation times to show that KPCA used in concert with LASSO classification is significantly faster than linear SVM over a wide range of training set sizes. Our work shows that powerful linear QSAR methods can match nonlinear methods and demonstrates a modular approach to nonlinear classification that greatly enhances QSAR model prototyping facility, flexibility, and transferability.

A mitochondrial inside-out iron-calcium signal reveals drug targets for Parkinson's disease.

Dysregulated iron or Ca2+ homeostasis has been reported in Parkinson's disease (PD) models. Here, we discover a connection between these two metals at the mitochondria. Elevation of iron levels causes inward mitochondrial Ca2+ overflow, through an interaction of Fe2+ with mitochondrial calcium uniporter (MCU). In PD neurons, iron accumulation-triggered Ca2+ influx across the mitochondrial surface leads to spatially confined Ca2+ elevation at the outer mitochondrial membrane, which is subsequently sensed by Miro1, a Ca2+-binding protein. A Miro1 blood test distinguishes PD patients from controls and responds to drug treatment. Miro1-based drug screens in PD cells discover Food and Drug Administration-approved T-type Ca2+-channel blockers. Human genetic analysis reveals enrichment of rare variants in T-type Ca2+-channel subtypes associated with PD status. Our results identify a molecular mechanism in PD pathophysiology and drug targets and candidates coupled with a convenient stratification method.

Application of MCAT questions as a testing tool and evaluation metric for knowledge graph-based reasoning systems.

"Knowledge graphs" (KGs) have become a common approach for representing biomedical knowledge. In a KG, multiple biomedical data sets can be linked together as a graph representation, with nodes representing entities, such as "chemical substance" or "genes," and edges representing predicates, such as "causes" or "treats." Reasoning and inference algorithms can then be applied to the KG and used to generate new knowledge. We developed three KG-based question-answering systems as part of the Biomedical Data Translator program. These systems are typically tested and evaluated using traditional software engineering tools and approaches. In this study, we explored a team-based approach to test and evaluate the prototype "Translator Reasoners" through the application of Medical College Admission Test (MCAT) questions. Specifically, we describe three "hackathons," in which the developers of each of the three systems worked together with a moderator to determine whether the applications could be used to solve MCAT questions. The results demonstrate progressive improvement in system performance, with 0% (0/5) correct answers during the first hackathon, 75% (3/4) correct during the second hackathon, and 100% (5/5) correct during the final hackathon. We discuss the technical and sociologic lessons learned and conclude that MCAT questions can be applied successfully in the context of moderated hackathons to test and evaluate prototype KG-based question-answering systems, identify gaps in current capabilities, and improve performance. Finally, we highlight several published clinical and translational science applications of the Translator Reasoners.

Machine learning in chemoinformatics and drug discovery.

Chemoinformatics is an established discipline focusing on extracting, processing and extrapolating meaningful data from chemical structures. With the rapid explosion of chemical 'big' data from HTS and combinatorial synthesis, machine learning has become an indispensable tool for drug designers to mine chemical information from large compound databases to design drugs with important biological properties. To process the chemical data, we first reviewed multiple processing layers in the chemoinformatics pipeline followed by the introduction of commonly used machine learning models in drug discovery and QSAR analysis. Here, we present basic principles and recent case studies to demonstrate the utility of machine learning techniques in chemoinformatics analyses; and we discuss limitations and future directions to guide further development in this evolving field.

Chemical reaction vector embeddings: towards predicting drug metabolism in the human gut microbiome.

Bacteria in the human gut have the ability to activate, inactivate, and reactivate drugs with both intended and unintended effects. For example, the drug digoxin is reduced to the inactive metabolite dihydrodigoxin by the gut Actinobacterium E. lenta, and patients colonized with high levels of drug metabolizing strains may have limited response to the drug. Understanding the complete space of drugs that are metabolized by the human gut microbiome is critical for predicting bacteria-drug relationships and their effects on individual patient response. Discovery and validation of drug metabolism via bacterial enzymes has yielded >50 drugs after nearly a century of experimental research. However, there are limited computational tools for screening drugs for potential metabolism by the gut microbiome. We developed a pipeline for comparing and characterizing chemical transformations using continuous vector representations of molecular structure learned using unsupervised representation learning. We applied this pipeline to chemical reaction data from MetaCyc to characterize the utility of vector representations for chemical reaction transformations. After clustering molecular and reaction vectors, we performed enrichment analyses and queries to characterize the space. We detected enriched enzyme names, Gene Ontology terms, and Enzyme Consortium (EC) classes within reaction clusters. In addition, we queried reactions against drug-metabolite transformations known to be metabolized by the human gut microbiome. The top results for these known drug transformations contained similar substructure modifications to the original drug pair. This work enables high throughput screening of drugs and their resulting metabolites against chemical reactions common to gut bacteria.