A Hybrid Imputation Method for Multi-Pattern Missing Data: A Case Study on Type II Diabetes Diagnosis

Real medical datasets usually consist of missing data with different patterns which decrease the performance of classifiers used in intelligent healthcare and disease diagnosis systems. Many methods have been proposed to impute missing data, however, they do not fulfill the need for data quality especially in real datasets with different missing data patterns. In this paper, a four-layer model is introduced, and then a hybrid imputation (HIMP) method using this model is proposed to impute multi-pattern missing data including non-random, random, and completely random patterns. In HIMP, first, non-random missing data patterns are imputed, and then the obtained dataset is decomposed into two datasets containing random and completely random missing data patterns. Then, concerning the missing data patterns in each dataset, different single or multiple imputation methods are used. Finally, the best-imputed datasets gained from random and completely random patterns are merged to form the final dataset. The experimental evaluation was conducted by a real dataset named IRDia including all three missing data patterns. The proposed method and comparative methods were compared using different classifiers in terms of accuracy, precision, recall, and F1-score. The classifiers’ performances show that the HIMP can impute multi-pattern missing values more effectively than other comparative methods.

Capillary-Like Formations of Endothelial Cells in Defined Patterns Generated by Laser Bioprinting

Bioprinting is seen as a promising technique for tissue engineering, with hopes of one day being able to produce whole organs. However, thick tissue requires a functional vascular network, which naturally contains vessels of various sizes, down to capillaries of ~10 µm in diameter, often spaced less than 200 µm apart. If such thick tissues are to be printed, the vasculature would likely need to be printed at the same time, including the capillaries. While there are many approaches in tissue engineering to produce larger vessels in a defined manner, the small capillaries usually arise only in random patterns by sprouting from the larger vessels or from randomly distributed endothelial cells. Here, we investigated whether the small capillaries could also be printed in predefined patterns. For this purpose, we used a laser-based bioprinting technique that allows for the combination of high resolution and high cell density. Our aim was to achieve the formation of closed tubular structures with lumina by laser-printed endothelial cells along the printed patterns on a surface and in bioprinted tissue. This study shows that such capillaries are directly printable; however, persistence of the printed tubular structures was achieved only in tissue with external stimulation by other cell types.

Physically unclonable security patterns created by electrospinning, and authenticated by two-step validation method

Counterfeiting is a growing economic and social problem. For anticounterfeiting, random and inimitable droplet/fiber patterns were created by the electrospinning method as security tags that are detectable under UV light but invisible in daylight. To check the authenticity of the original security patterns created; images were collected with a simple smartphone microscope and a database of the recorded original patterns was created. The originality of the random patterns was checked by comparing them with the patterns recorded in the database. In addition, the spectral signature of the patterns in the droplet/fiber network was obtained with a simple and hand-held spectrometer. Thus, by reading the spectral signature from the pattern, the spectral information of the photoluminescent nanoparticles was verified and thus a second-step verification was established. In this way, anticounterfeiting technology that combines ink formula, unclonable security pattern creation and two-level verification is developed.

Live Analysis of Position Effect Variegation Reveals Different Modes of Action for HP1 and Su(var)3-9

Position Effect Variegation (PEV) results from the juxtaposition of euchromatic and heterochromatic components of eukaryotic genomes, silencing genes near the new euchromatin/heterochromatin junctions. The degree of silencing is itself heritable through S-phase, giving rise to distinctive random patterns of cell clones expressing the genes intermixed with clones in which the genes are silenced. Much of what we know about epigenetic inheritance in the soma stems from work on PEV aimed at identifying the components of the silencing machinery and its mechanism of inheritance. Despite identifying two central gene activities - the Su(var)3-9 histone H3-Lysine-9 methyltransferase and the Su(var)205/HP1 methyl-H3-Lysine-9 binding protein - their role in PEV has been inferred from terminal phenotypes, leaving considerable gaps in understanding how PEV behaves through development. Here, we investigate the phenotypes of Su(var)3-9 and Su(var)205/HP1 mutations in live developing tissues. We discovered that mutations in Su(var)205/HP1 compromise the initial establishment of PEV in early embryogenesis. Later gains of heterochromatin-induced gene silencing are possible, but are unstable and lost rapidly. In contrast, mutations in Su(var)3-9 exhibit robust silencing early in development, but fail to maintain it through subsequent cell divisions. Our analyses show that while the terminal phenotypes of these mutations may appear identical, they have arrived at them through different developmental trajectories. We discuss how our findings further challenge existing models for epigenetic inheritance of heterochromatin-induced gene silencing.

Learning from the resourceness blind spot for service innovation at the base of the pyramid

Purpose This paper aims to investigate whether and why the base of the pyramid (BOP) actors display passive innovation resistance because of which they reject service innovations without evaluation and forfeit potential to improve their well-being. The resourceness concept, referring to the outcome of how actors appraise and integrate resources in pursuit of a purpose at hand, is used as a theoretical lens to investigate the everyday consumption behaviour of BOP households and helps to investigate how and why passive innovation resistance occurs. The outcomes of the study help address important theoretical and practical considerations for the development of successful new service concepts at the BOP. Design/methodology/approach Narrative interviews with 29 households in Zambia provide data, from which patterns in how potential resources do or do not become real are identified and related to the concept of passive innovation. Findings Economic, social and other factors in the BOP context clearly influence non-random patterns of resource integration which are correlated with passive innovation resistance. This can lead to service innovations being ignored and/or misunderstood prior to evaluation for adoption. This is a risk to the potential positive impact of service innovation for poverty alleviation at the BOP. Practical implications Service innovation at the BOP must begin with a deep understanding of “how” and “why” consumers typically appraise and integrate potential resources to achieve a beneficial outcome in their context. To overcome the barrier of passive innovation resistance, marketing education must stimulate an understanding of potential benefits and motivation towards the change associated with the adoption of service innovation. Social implications The findings support more successful service innovation strategies for the BOP, which can provide vital infrastructure for the alleviation of poverty. Originality/value The application of a service-dominant logic perspective in the BOP context and the conceptual linkage between resourceness and passive innovation resistance is novel. Valuable insights are gained for service practitioners at the BOP and for further conceptual development of innovation resistance in the BOP context.

End-to-End Control Chart Pattern Classification Using a 1D Convolutional Neural Network and Transfer Learning

Control charts are an important tool in statistical process control (SPC). They have been commonly used for monitoring process variation in many industries. Recognition of non-random patterns is an important task in SPC. The presence of non-random patterns implies that a process is affected by certain assignable causes, and some corrective actions should be taken. In recent years, a great deal of research has been devoted to the application of machine learning (ML) based approaches to control chart pattern recognition (CCPR). However, there are some gaps that hinder the application of the CCPR methods in practice. In this study, we applied a control chart pattern recognition method based on an end-to-end one-dimensional convolutional neural network (1D CNN) model. We proposed some methods to generate datasets with high intra-class diversity aiming to create a robust classification model. To address the data scarcity issue, some data augmentation operations suitable for CCPR were proposed. This study also investigated the usefulness of transfer learning techniques for the CCPR task. The pre-trained model using normally distributed data was used as a starting point and fine-tuned on the unknown non-normal data. The performance of the proposed approach was evaluated by real-world data and simulation experiments. Experimental results indicate that our proposed method outperforms the traditional machine learning methods and could be a promising tool to effectively classify control chart patterns. The results and findings of this study are crucial for the further realization of smart statistical process control.

Portfolio Correlations in the Bank-Firm Credit Market of Japan

The recent global financial crisis has shown portfolio correlations between agents as one of the major channels of risk contagion and amplification. In this work, we analyse the structure and dynamics of the cross-correlation matrix of banks’ loan portfolios in the yearly bank-firm credit network of Japan during the period from 1980 to 2012. Using the methods of Random Matrix Theory (RMT), Principal Component Analysis and complex networks, we aim to detect non-random patterns in the empirical cross-correlations as well as to identify different states of such correlations over time. Our findings suggest that although a majority of portfolio correlations between banks in lending relations to firms are contributed by noise, the top largest eigenvalues always deviate from the random bulk explained by RMT, indicating the presence of non-random patterns governing the correlation dynamics. In particular, we show that this dynamics is mainly driven by a global common factor and a couple of “groups” factors. Furthermore, different states in the credit market can be identified based on the evolution of eigenvalues and associated eigenvectors. For example, during the asset price bubble period in Japan from 1986 to 1991, we find that banks’ loan portfolios tend to be more correlated, showing a significant increase in the level of systemic risk in the credit market. In addition, building Planar Maximally Filtered Graphs from the correlations of different eigenmodes, notably, we observe that the local interaction structure between banks changes in different periods. Typically, when the dominance of a group of banks in one period gradually vanishes, the credit market starts to build-up a different structure in the next period in which another group of banks will become the main actors in the backbone of the cross-correlations.

Effect of Non-canonical Spatial Symmetry on Subitizing

Subitizing refers to ability of people to accurately and effortlessly enumerate a small number of items, with a capacity around four elements. Previous research showed that “canonical” organizations, such as familiar layouts on a dice, can readily improve subitizing performance of people. However, almost all canonical shapes found in the world are also highly symmetrical; therefore, it is unclear whether previously reported facilitative effect of canonical organization is really due to canonicality, or simply driven by spatial symmetry. Here, we investigated the possible effect of symmetry on subitizing by using symmetrical, yet non-canonical, shape structures. These symmetrical layouts were compared with highly controlled random patterns (Experiment 1), as well as fully random and canonical patterns (Experiment 2). Our results showed that symmetry facilitates subitizing performance, but only at set size of 6, suggesting that the effect is insufficient to improve performance of people in the lower or upper range. This was also true, although weaker, in reaction time (RT), error distance measures, and Weber Fractions. On the other hand, canonical layouts produced faster and more accurate subitizing performances across multiple set sizes. We conclude that, although previous findings mixed symmetry in their canonical shapes, their findings on shape canonicality cannot be explained by symmetry alone. We also propose that our symmetrical and canonical results are best explained by the “groupitizing” and pattern recognition accounts, respectively.