All content is freely available to readers and supported through open access

Jan 14, 2022

Volume 3Issue 1
Open Access
On the cover: In Volinski et al. (100391), we extend the traditional discussion of conventional and geometrically constrained inverse kinematics in brain-inspired (neuromorphic) robotics to learning-based and deep spiking neural networks. The cover illustrates our neuromorphically controlled robotic arm, where spikes, demonstrated with a raster plot, are visualized in the background. Image credit: Taylor Tabb, Accenture Labs, San Francisco....
On the cover: In Volinski et al. (100391), we extend the traditional discussion of conventional and geometrically constrained inverse kinematics in brain-inspired (neuromorphic) robotics to learning-based and deep spiking neural networks. The cover illustrates our neuromorphically controlled robotic arm, where spikes, demonstrated with a raster plot, are visualized in the background. Image credit: Taylor Tabb, Accenture Labs, San Francisco.


People of data

  • Walking on the thin intersectional lines of disciplines

    • Elishai Ezra Tsur,
    • Travis DeWolf,
    • Lazar Supic
    Elishai Ezra Tsur, a multidisciplinary researcher, talks about the challenges that conventional academic mindset brought to his professional life. He, DeWolf, and Supic introduce us with their viewpoint about “data science” and its role in their research. In their recent work published in this issue of Patterns, they tackle the inverse kinematics problem using brain-inspired neuronal architectures.


  • African genomic data sharing and the struggle for equitable benefit

    • Michèle Ramsay
    Genomic and related health data from Africa remain scarce and are extremely valuable, due to an abundance of variants often rare or absent in the rest of the world. Insights from such data will benefit global populations, but will Africa be neglected by limited access to affordable benefits resulting from research that uses their data?


  • In medicine, how do we machine learn anything real?

    • Marzyeh Ghassemi,
    • Elaine Okanyene Nsoesie
    In this article, we provide examples of bias in clinical devices, interventions, and interactions to highlight the pervasiveness of inequity in health data and elaborate on the potential of these data to repeat or worsen inequity if used to develop machine learning algorithms. To address these issues, individuals and institutions must recognize implicit bias, adopt a “do-no-harm” approach to applying algorithms, implement systemic anti-racist policies and practices, and adopt approaches that have been introduced in other fields.
  • Blockchain humanitarianism and crypto-colonialism

    • Olivier Jutel
    This paper examines the ability of blockchain developers to use the developing world for experimentation as a result of the public-private partnership model. There is evidence of a revolving door between the NGO sector and the start-up space. Blockchain humanitarian projects have thus been central to the PR hype around crypto and are an indispensable form of legitimation. A lack of proper ethical boundaries is evident in blockchain humanitarian projects that exist alongside some of the most dubious elements of the crypto-economy.


  • The impact universe—a framework for prioritizing the public interest in the Internet of Things

    • Francine Berman,
    • Emilia Cabrera,
    • Ali Jebari,
    • Wassim Marrakchi
    The connected technologies of the Internet of Things (IoT) drive the world we live in and have profound implications for society. When humans are empowered by technology and technology learns from experience, a new kind of social contract is needed, one that recognizes the benefits, risks, and broad impacts of innovation. Development of an “impact universe” can holistically expose potential IoT outcomes and guide socially responsible design, use, and oversight of IoT systems that advance the public interest.


  • Perception of fairness in algorithmic decisions: Future developers' perspective

    • Styliani Kleanthous,
    • Maria Kasinidou,
    • Pınar Barlas,
    • Jahna Otterbacher
    The importance of this work lays primarily in that, while others have looked into how the end users and/or the general public perceive elements of fairness, accountability, rransparency, and ethics, it is important to understand how the people who are involved in the development of algorithmic decision-making systems perceive the above concepts. We approach this through a study with an international sample of students coming from fields adjacent to computing in order to examine their perceptions on the above topics.
  • Machine learning for predicting readmission risk among the frail: Explainable AI for healthcare

    • Somya D. Mohanty,
    • Deborah Lekan,
    • Thomas P. McCoy,
    • Marjorie Jenkins,
    • Prashanti Manda
    Automated prediction of readmission risk has the potential to save millions of dollars in healthcare costs and can improve patient care. Our work presents machine learning models that take into account various facets of patients, such as demographics, comorbidities, and frailty parameters, to accurately estimate their risk of being readmitted within 30 days. We place high importance on explainability, thereby enhancing confidence in the automated models.
  • Machine learning and network medicine approaches for drug repositioning for COVID-19

    • Suzana de Siqueira Santos,
    • Mateo Torres,
    • Diego Galeano,
    • María del Mar Sánchez,
    • Luca Cernuzzi,
    • Alberto Paccanaro
    We present two complementary machine learning approaches for drug repositioning against COVID-19 that target SARS-CoV-2 and its cellular processes in the host, respectively. Our matrix decomposition approach exploits drug developmental information to predict broad-spectrum antivirals; our graph kernel-based approach, rooted in ideas from network medicine, predicts which FDA-approved drugs are more likely to perturb the human subnetwork that is crucial for SARS-CoV-2 infection/replication. We also introduce CoREx, a freely available online tool to reason and formulate hypothesis about drug repurposing in the context of biological networks and pharmacological information.
  • Contrastive learning of heart and lung sounds for label-efficient diagnosis

    • Pratham N. Soni,
    • Siyu Shi,
    • Pranav R. Sriram,
    • Andrew Y. Ng,
    • Pranav Rajpurkar
    The project focuses on a common problem in machine learning in specialized fields: a lack of labeled data. We propose a solution with a contrastive learning framework to leverage unlabeled data and use associated metadata for pair selection to learn data representations. With this framework, we are able to improve downstream tasks performances with limited labeled data. We choose to apply this framework in healthcare, where unlabeled data are abundant and data labeling is expensive.
  • Data-driven artificial and spiking neural networks for inverse kinematics in neurorobotics

    • Alex Volinski,
    • Yuval Zaidel,
    • Albert Shalumov,
    • Travis DeWolf,
    • Lazar Supic,
    • Elishai Ezra Tsur
    We demonstrate a data-driven approach for robotic motion planning in complex environments, relying on the versatility of neural networks. We extend the discussion to brain-inspired neuronal architectures, where spiking neural networks constitute the computational framework. Spiking neural networks underlie the field of neurorobotics. They are argued to enable a robotic control that outperforms conventional paradigms in terms of robustness to perturbations and adaptation to varying conditions. A brain-inspired efficient implementation of inverse kinematics is, therefore, an important stepping stone in neurorobotics.
  • Prediction of allosteric sites and signaling: Insights from benchmarking datasets

    • Nan Wu,
    • Léonie Strömich,
    • Sophia N. Yaliraki
    Modulation of protein activity is important for disease treatment, and challenges associated with common drugs have fueled the need for allosteric drugs that bind distant from the active site. Yet, allosteric site prediction remains challenging. In this study, we benchmark an atomistic, graph-theoretical method called bond-to-bond propensity against allosteric protein databases to illustrate its capability to accurately uncover allosteric sites. Our six scoring measures provide additional insights into potential mechanisms involved and could guide the design of drug molecules.
  • Auto-annotating sleep stages based on polysomnographic data

    • Hanrui Zhang,
    • Xueqing Wang,
    • Hongyang Li,
    • Soham Mehendale,
    • Yuanfang Guan
    In this study, we developed an automatic and fast sleeping stage and arousal/apnea detection tool, by adapting a U-net architecture with a convolutional neural network that is suitable for processing temporal information and makes sequence-to-sequence annotations. Our model is tested on different modalities and is consistently achieving excellent performance, which is comparable with human experts. Our tool provides an alternative to assist human experts in detecting pathological sleeping patterns in the study of clinical patients.