Search

The second Machine Learning for Health (ML4H) symposium was held both virtually and in-person on November 28, 2022, in New Orleans, Louisiana, USA (Parziale et al.,2022). The symposium included research roundtable sessions to foster discussions between participants and senior researchers on timely and relevant topics for the ML4H community. Encouraged by the successful virtual roundtables in the previous year (Roy et al.,2021), we organized nine in-person and four virtual roundtables at ML4H 2022 (Parziale et al.,2022). A roundtable session included invited senior chairs (with substantial experience in the field), junior chairs (responsible for facilitating the discussion), and attendees from di- verse backgrounds with interest in the session’s topic. This document explains the organization process we used and compiles the takeaways from the roundtable discussions, including recent advances, applications, and open challenges for each topic. We conclude with a summary and lessons learned across all roundtables. more...

Health insurers often use algorithms to identify members who would benefit from care and condition management programs, which provide personalized, high-touch clinical support. Timely, accurate, and seamless integration between algorithmic identification and clinical intervention depends on effective collaboration between the system designers and nurse care managers. We focus on a high-risk pregnancy (HRP) program designed to reduce the likelihood of adverse prenatal, perinatal, and postnatal events and describe how we overcome three challenges of HRP programs as articulated by nurse care managers; (1) early detection of pregnancy, (2) accurate identification of impactable high-risk members, and (3) provision of explainable indicators to supplement predictions. We propose a novel algorithm for pregnancy identification that identifies pregnancies 57 days earlier than previous code-based models in a retrospective study. We then build a model to predict impactable pregnancy complications that achieves an AUROC of 0.760. Models for pregnancy identification and complications are then integrated into a proposed user interface. In a set of user studies, we collected quantitative and qualitative feedback from nurses on the utility of the predictions combined with clinical information driving the predictions on triaging members for the HRP program. more...

SARS-CoV-2 Delta and Omicron strains are the most globally relevant variants of concern (VOCs). While individuals infected with Delta are at risk to develop severe lung disease1, Omicron infection causes less severe disease, mostly upper respiratory symptoms2,3. The question arises whether rampant spread of Omicron could lead to mass immunization, accelerating the end of the pandemic. Here we show that infection with Delta, but not Omicron, induces broad immunity in mice. While sera from Omicron-infected mice only neutralize Omicron, sera from Delta-infected mice are broadly effective against Delta and other VOCs, including Omicron. This is not observed with the WA1 ancestral strain, although both WA1 and Delta elicited a highly pro-inflammatory cytokine response and replicated to similar titers in the respiratory tracts and lungs of infected mice as well as in human airway organoids. Pulmonary viral replication, pro-inflammatory cytokine expression, and overall disease progression are markedly reduced with Omicron infection. Analysis of human sera from Omicron and Delta breakthrough cases reveals effective cross-variant neutralization induced by both viruses in vaccinated individuals. Together, our results indicate that Omicron infection enhances preexisting immunity elicited by vaccines, but on its own may not induce broad, cross-neutralizing humoral immunity in unvaccinated individuals. more...

Efforts to determine why new severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants demonstrate improved fitness have been limited to analyzing mutations in the spike (S) protein with the use of S-pseudotyped particles. In this study, we show that SARS-CoV-2 virus-like particles (SC2-VLPs) can package and deliver exogenous transcripts, enabling analysis of mutations within all structural proteins and at multiple steps in the viral life cycle. In SC2-VLPs, four nucleocapsid (N) mutations found universally in more-transmissible variants independently increased messenger RNA delivery and expression ~10-fold, and in a reverse genetics model, the serine-202→arginine (S202R) and arginine-203→methionine (R203M) mutations each produced >50 times as much virus. SC2-VLPs provide a platform for rapid testing of viral variants outside of a biosafety level 3 setting and demonstrate N mutations and particle assembly to be mechanisms that could explain the increased spread of variants, including B.1.617.2 (Delta, which contains the R203M mutation). more...

Unsupervised learning is often used to uncover clusters in data. However, different kinds of noise may impede the discovery of useful patterns from real-world time-series data. In this work, we focus on mitigating the interference of interval censoring in the task of clustering for disease phenotyping. We develop a deep generative, continuous-time model of time-series data that clusters time-series while correcting for censorship time. We provide conditions under which clusters and the amount of delayed entry may be identified from data under a noiseless model. On synthetic data, we demonstrate accurate, stable, and interpretable results that outperform several benchmarks. On real-world clinical datasets of heart failure and Parkinson's disease patients, we study how interval censoring can adversely affect the task of disease phenotyping. Our model corrects for this source of error and recovers known clinical subtypes., Comment: AAAI 2022 more...

The ability of enzymes, including ribozymes, to catalyze side reactions is believed to be essential to the evolution of novel biochemical activities. It has been speculated that the earliest ribozymes, whose emergence marked the origin of life, were low in activity but high in promiscuity, and that these early ribozymes gave rise to specialized descendants with higher activity and specificity. Here, we review the concepts related to promiscuity and examine several cases of highly promiscuous ribozymes. We consider the evidence bearing on the question of whether de novo ribozymes would be quantitatively more promiscuous than later evolved ribozymes or protein enzymes. We suggest that while de novo ribozymes appear to be promiscuous in general, they are not obviously more promiscuous than more highly evolved or active sequences. Promiscuity is a trait whose value would depend on selective pressures, even during prebiotic evolution. more...

In vitro selection using mRNA display is currently a widely used method to isolate functional peptides with desired properties. The analysis of high throughput sequencing (HTS) data from in vitro evolution experiments has proven to be a powerful technique but only recently has it been applied to mRNA display selections. In this Perspective, we introduce aspects of mRNA display and HTS that may be of interest to physical chemists. We highlight the potential of HTS to analyze in vitro selections of peptides and review recent advances in the application of HTS analysis to mRNA display experiments. We discuss some possible issues involved with HTS analysis and summarize some strategies to alleviate them. Finally, the potential for future impact of advancing HTS analysis on mRNA display experiments is discussed. more...

This volume represents the accepted submissions from the Machine Learning for Health (ML4H) workshop at the conference on Neural Information Processing Systems (NeurIPS) 2018, held on December 8, 2018 in Montreal, Canada., Comment: Machine Learning for Health (ML4H) Workshop at NeurIPS 2018 arXiv:1811.07216 more...

In vitro selection experiments in biochemistry allow for the discovery of novel molecules capable of specific desired biochemical functions. However, this is not the only benefit we can obtain from such selection experiments. Since selection from a random library yields an unprecedented, and sometimes comprehensive, view of how a particular biochemical function is distributed across sequence space, selection experiments also provide data for creating and analyzing molecular fitness landscapes, which directly map function (phenotypes) to sequence information (genotypes). Given the importance of understanding the relationship between sequence and functional activity, reliable methods to build and analyze fitness landscapes are needed. Here, we present some statistical methods to extract this information from pools of RNA molecules. We also provide new computational tools to construct and study molecular fitness landscapes. more...

Designs derived from nature have become a perfect blueprint for today's engineers and scientists to follow and implement. One particularly noted area is the defense industry, wherein flexible dermal armor inspired by nature has been pioneering many sophisticated technologies and designs in recent years. Designers today are considering borrowing aspects of flexibility and mobility of natural dermal armors to enhance the maneuverability of man-made armor by imitating the following mechanisms : (1) 3D interlocking; (2) interconnectivity; (3) hierarchical structure; (4) structural toughening mechanisms. The motive of this research is to examine the flexible dermal armors that Mother Nature has offered. The innate structure and mechanical properties of osteoderms in the following animals have been investigated: armadillo, alligator, and leatherback sea turtle. All these animals have one commonality: they have bony plates embedded in their soft shell that make their skin flexible and compliant while also offering protection. Each of these animals has its own unique way with associated functional mechanisms to achieve this. First, the armadillo uses a flexible biological material, collagen, to connect the bony plates and to provide flexibility. Second, the osteoderms of an alligator have bone-toughening mechanisms to reinforce themselves from a predator's attack. Third, the leatherback sea turtle exhibits a zigzag suture design that incorporates interlocking methods into its armor that enables it to dive deep into the ocean (depth of 1200 meters). Mechanical tests, characterization, and analysis were performed to identify the principal mechanisms that provide both toughness and flexibility in the three different dermal armors. The damage mechanisms are connected to the microstructural characteristics. The toughness and flexibility of the natural designs were found to be quite remarkable more...

Carnegie Mellon Silicon Valley

Figure S1. Graphical schematic of fractionation method. Figure S2. Additional comparisons of VLP-enriched DNA composition from clinical skin and wound swabs using kit-based extraction (PS1) and the method described here (PS2). VLP-enriched DNA was shotgun sequenced and mapped to the IMG/VR viral metagenome database (A) or the human genome (B). Means are compared by t-tests. Although PS2 samples contained more human DNA than PS1, the greater fraction of viral reads (Fig. 6b) also translated to a greater absolute number of reads mapping to IMG/VR. (PDF 702 kb) more...

Figure S1. Graphical schematic of fractionation method. Figure S2. Additional comparisons of VLP-enriched DNA composition from clinical skin and wound swabs using kit-based extraction (PS1) and the method described here (PS2). VLP-enriched DNA was shotgun sequenced and mapped to the IMG/VR viral metagenome database (A) or the human genome (B). Means are compared by t-tests. Although PS2 samples contained more human DNA than PS1, the greater fraction of viral reads (Fig. 6b) also translated to a greater absolute number of reads mapping to IMG/VR. (PDF 702 kb) more...

This chapter explores the extent to which selected writings of French philosopher Michel Serres and a health care model created by Brian Hodges in the UK can augment and inform the development of social informatics. The volume of Serres' output contrasts markedly with work devoted to Hodges' Health Career - Care Domains – Model. Since the concept of health is universal culturally, and informatics disciplines are emerging fields of practice characterised by indistinct boundaries in terms of theory, policy and practice, various ethnographic and cultural associations will be made. Placing Hodges’ model and Serres’ work together is not intended to suggest direct equivalence, other than the common themes this author intends to bring to the attention of the social informatics community. Central to the above, is the notion of holistic bandwidth, utilising Hodges’ model as a tool to develop and disseminate socio-technical perspectives. more...