Citations or dollars? Early signals of a firm's research success

Scientific and technological progress is largely driven by firms in many domains, including artificial intelligence and vaccine development. However, we do not know yet whether the success of firms’ research activities exhibits dynamic regularities and some degree of predictability. By inspecting the research lifecycles of 7,440 publicly listed firms, we find that the economic value of a firm’s early patents is an accurate predictor of various dimensions of a firm’s future research success. At the same time, a smaller set of future top-performers do not generate early patents of high economic value, but they are detectable via the technological value of their early patents. Importantly, the observed predictability cannot be explained by a cumulative advantage mechanism, and the observed heterogeneity of the firms’ temporal success patterns markedly differs from patterns previously observed for individuals’ research careers. Our results uncover the dynamical regularities of the research success of firms, and they could inform managerial strategies as well as policies to promote corporate development and accelerate human progress.

Best practices in the performance of District Hospitals in India

NITI Aayog undertook this novel exercise on the assessment of the performance of District Hospitals across the country for the financial year 2017-18 in collaboration with the Ministry of Health and Family Welfare, NABH and WHO Country Office for India. The assessment framework covers 10 Key Performance Indicators (KPIs) across the domains of Structure and Output. A total of 707 district hospitals across all States and Union Territories participated in the performance assessment. The framework classifies hospitals in three categories namely Small Hospitals (less than or equal to 200 beds), Mid-sized Hospitals (between 201-300 beds) and Large Hospitals (more than 300 beds). Three top-performing district hospitals in each of the hospital categories across the 10 KPIs were identified and their best practices collected and documented. In all, 75 district hospitals across 24 States and Union Territories emerged as top performers on indicators ranging from availability of beds, medical and paramedical staff, core health and diagnostic testing services to outputs such as bed occupancy rate and number of surgeries per surgeon etc. We believe that this country-wide assessment of district hospitals will serve as a valuable resource for them to learn best practices from each other in a spirit to improve their performance. It will also provide an opportunity to showcase progress.

Tree-based machine learning performed in-memory with memristive analog CAM

Tree-based machine learning techniques, such as Decision Trees and Random Forests, are top performers in several domains as they do well with limited training datasets and offer improved interpretability compared to Deep Neural Networks (DNN). However, these models are difficult to optimize for fast inference at scale without accuracy loss in von Neumann architectures due to non-uniform memory access patterns. Recently, we proposed a novel analog content addressable memory (CAM) based on emerging memristor devices for fast look-up table operations. Here, we propose for the first time to use the analog CAM as an in-memory computational primitive to accelerate tree-based model inference. We demonstrate an efficient mapping algorithm leveraging the new analog CAM capabilities such that each root to leaf path of a Decision Tree is programmed into a row. This new in-memory compute concept for enables few-cycle model inference, dramatically increasing 103 × the throughput over conventional approaches.

Sequence-Based Explainable Hybrid Song Recommendation

Despite advances in deep learning methods for song recommendation, most existing methods do not take advantage of the sequential nature of song content. In addition, there is a lack of methods that can explain their predictions using the content of recommended songs and only a few approaches can handle the item cold start problem. In this work, we propose a hybrid deep learning model that uses collaborative filtering (CF) and deep learning sequence models on the Musical Instrument Digital Interface (MIDI) content of songs to provide accurate recommendations, while also being able to generate a relevant, personalized explanation for each recommended song. Compared to state-of-the-art methods, our validation experiments showed that in addition to generating explainable recommendations, our model stood out among the top performers in terms of recommendation accuracy and the ability to handle the item cold start problem. Moreover, validation shows that our personalized explanations capture properties that are in accordance with the user’s preferences.

Critical Assessment of Metagenome Interpretation - the second round of challenges

Evaluating metagenomic software is key for optimizing metagenome interpretation and focus of the community-driven initiative for the Critical Assessment of Metagenome Interpretation (CAMI). In its second challenge, CAMI engaged the community to assess their methods on realistic and complex metagenomic datasets with long and short reads, created from ~1,700 novel and known microbial genomes, as well as ~600 novel plasmids and viruses. Altogether 5,002 results by 76 program versions were analyzed, representing a 22x increase in results. Substantial improvements were seen in metagenome assembly, some due to using long-read data. The presence of related strains still was challenging for assembly and genome binning, as was assembly quality for the latter. Taxon profilers demonstrated a marked maturation, with taxon profilers and binners excelling at higher bacterial taxonomic ranks, but underperforming for viruses and archaea. Assessment of clinical pathogen detection techniques revealed a need to improve reproducibility. Analysis of program runtimes and memory usage identified highly efficient programs, including some top performers with other metrics. The CAMI II results identify current challenges, but also guide researchers in selecting methods for specific analyses.

Voting in multi-stage elimination contests: Evidence from a Karaoke show

We employ the data from a karaoke contest to analyze strategic voting. Participants face a trade-off when voting for the contestant they want to eliminate. Excluding worst-performers increases the size of the prize allocated to the winner, whereas excluding top-performers increases the chances to become the winner. We analyze the performance and voting decisions and justifications of 138 participants in this contest across 23 episodes. We find that votes for worst-performers are much more common than votes for top-performers, and the justifications for voting due to the competitors' mistakes are the most prominent. Although contestants are not informed of the performance of themselves or any other participant, the likelihood to vote for the worst-performer is higher than the probability of randomly voting for someone else.

Should policy makers trust composite indices? A commentary on the pitfalls of inappropriate indices for policy formation

Background This paper critically discusses the use and merits of global indices, in particular, the Global Health Security Index (GHSI; Cameron et al. https://www.ghsindex.org/#l-section--map) in times of an imminent crisis, such as the current pandemic. This index ranked 195 countries according to their expected preparedness in the case of a pandemic or other biological threat. The coronavirus disease 2019 (Covid-19) pandemic provides the background to compare each country's predicted performance from the GHSI with the actual performance. In general, there is an inverted relation between predicted versus actual performance, i.e. the predicted top performers are among those that are the worst hit. Obviously, this reflects poorly on the potential policy uses of this index in imminent crisis management. Methods The paper analyses the GHSI and identifies why it may have struggled to predict actual pandemic preparedness as evidenced by the Covid-19 pandemic. The paper also uses two different data sets, one from the Worldmeter on the spread of the Covid-19 pandemics, and the other from the International Network for Government Science Advice (INGSA) Evidence-to-Policy Tracker, to draw comparisons between the actual introduction of pandemic response policies and the corresponding death rate in 29 selected countries. Results This paper analyses the reasons for the poor match between prediction and reality in the index, and mentions six general observations applying to global indices in this respect. These observations are based on methodological and conceptual analyses. The level of abstraction in these global indices builds uncertainties upon uncertainties and hides implicit value assumptions, which potentially removes them from the policy needs on the ground. Conclusions From the analysis, the question is raised if the policy community might have better tools for decision-making in a pandemic. On the basis of data from the INGSA Evidence-to-Policy Tracker, and with backing in studies from social psychology and philosophy of science, some simple heuristics are suggested, which may be more useful than a global index.

Perylene diimide based non-fullerene acceptors: top performers and an emerging class featuring N-annulation

This perspective showcases new materials designs for perylene diimide based non-fullerene acceptors towards high performance photovoltaic devices.

Peak Age and Relative Performance Progression in International Cross-Country Skiers

Purpose: To quantify peak age and relative performance progression toward peak age in cross-country skiing according to event type, sex, and athlete performance level. Methods: International Ski Federation (FIS) points (performance expressed relative to the best athlete) of athletes born between 1981 and 1991, competing in junior world championships or finishing top 30 in world championships or Olympics, were downloaded from the FIS website. Individual performance trends were derived by fitting a quadratic curve to each athletes FIS point and age data. Results: Peak age was 26.2 (2.3) years in distance and 26.0 (1.7) years in sprint events. The sex difference in peak age in sprint events was ∼0.8 years (small, P = .001), while there was no significant sex difference in peak age in distance events (P = .668). Top performers displayed higher peak ages than other athletes in distance (mean difference, ±95% confidence limits = 1.6 y, ±0.6 y, moderate, P < .001) and sprint events (1.0, ±0.6 y, moderate, P < .001). FIS point improvement over the 5 years preceding peak age did not differ between event types (P = .325), while men improved more than women in both events (8.8, ±5.4%, small, P = .002 and 7.5, ±6.4%, small, P = .002). Performance level had a large effect on improvement in FIS points in both events (P < .001). Conclusion: This study provides novel insights on peak age and relative performance progression among world-class cross-country skiers and can assist practitioners, sport institutions, and federations with goal setting and evaluating strategies for achieving success.