Near-perfect Precision Research Articles

Optimising Human-Machine Collaboration for Efficient High-Precision Information Extraction from Text Documents

From science to law enforcement, many research questions are answerable only by poring over a large amount of unstructured text documents. While people can extract information from such documents with high accuracy, this is often too time-consuming to be practical. On the other hand, automated approaches produce nearly-immediate results, but are not reliable enough for applications where near-perfect precision is essential. Motivated by two use cases from criminal justice, we consider the benefits and drawbacks of various human-only, human–machine, and machine-only approaches. Finding no tool well suited for our use cases, we develop a human-in-the-loop method for fast but accurate extraction of structured data from unstructured text. The tool is based on automated extraction followed by human validation, and is particularly useful in cases where purely manual extraction is not practical. Testing on three criminal justice datasets, we find that the combination of the computer speed and human understanding yields precision comparable to manual annotation while requiring only a fraction of time, and significantly outperforms the precision of all fully automated baselines.

A Step Towards Automated Haematology: DL Models for Blood Cell Detection and Classification

INTRODUCTION: Deep Learning has significantly impacted various domains, including medical imaging and diagnostics, by enabling accurate classification tasks. This research focuses on leveraging deep learning models to automate the classification of different blood cell types, thus advancing hematology practices.
 OBJECTIVES: The primary objective of this study is to evaluate the performance of five deep learning models - ResNet50, AlexNet, MobileNetV2, VGG16, and VGG19 - in accurately discerning and classifying distinct blood cell categories: Eosinophils, Lymphocytes, Monocytes, and Neutrophils. The study aims to identify the most effective model for automating hematology processes.
 METHODS: A comprehensive dataset containing approximately 8,500 augmented images of the four blood cell types is utilized for training and evaluation. The deep learning models undergo extensive training using this dataset. Performance assessment is conducted using various metrics including accuracy, precision, recall, and F1-score.
 RESULTS: The VGG19 model emerges as the top performer, achieving an impressive accuracy of 99% with near-perfect precision and recall across all cell types. This indicates its robustness and effectiveness in automated blood cell classification tasks. Other models, while demonstrating competence, do not match the performance levels attained by VGG19.
 CONCLUSION: This research underscores the potential of deep learning in automating and enhancing the accuracy of blood cell classification, thereby addressing the labor-intensive and error-prone nature of traditional methods in hematology. The superiority of the VGG19 model highlights its suitability for practical implementation in real-world scenarios. However, further investigation is warranted to comprehend model performance variations and ensure generalization to unseen data. Overall, this study serves as a crucial step towards broader applications of artificial intelligence in medical diagnostics, particularly in the realm of automated hematology, fostering advancements in healthcare technology.

Investigating image-based fallow weed detection performance on Raphanus sativus and Avena sativa at speeds up to 30 km h−1

Site-specific weed control (SSWC) can provide considerable reductions in weed control costs and herbicide usage. Despite the promise of machine vision for SSWC systems and the importance of ground speed in weed control efficacy, there has been little investigation of the role of ground speed and camera characteristics on weed detection performance. Here, we compare the performance of four camera/software combinations using the open-source OpenWeedLocator platform – (1) default settings on a Raspberry Pi HQ camera (HQ1); (2) optimised software settings on a HQ camera (HQ2), (3) optimised software settings on the Raspberry Pi v2 camera (V2); and (4) a global shutter Arducam AR0234 camera (ARD) – at speeds ranging from 5 km h−1 to 30 km h−1. A combined excess green (ExG) and hue, saturation, value (HSV) thresholding algorithm was used for testing under fallow conditions using tillage radish (Raphanus sativus) and forage oats (Avena sativa) as representative broadleaf and grass weeds, respectively. ARD demonstrated the highest recall among camera systems, with up to 100 % recall for tillage radish and 91.6 % for forage oats at 5 km h−1. At 30 km h−1 ARD recall for forage oats declined (P < 0.05) by 12.4 %. There was no significant change for tillage radish. All cameras experienced a decrease in recall as speed increased. The highest rate of decrease was observed for HQ1 with 1.12 % and 0.90 % reductions in recall for every km/h increase in speed for tillage radish and forage oats, respectively. Detection of the ‘grassy’ forage oats was worse (P < 0.05) than the broadleaved tillage radish for all cameras. Despite the variations in recall, HQ1, HQ2, and V2 maintained near-perfect precision at all tested speeds. The variable effect of ground speed and camera system on detection performance of grass and broadleaf weeds, indicates that careful hardware and software considerations must be made when developing SSWC systems.

Beyond the Hype: An Evaluation of Commercially Available Machine Learning–based Malware Detectors

There is a lack of scientific testing of commercially available malware detectors, especially those that boast accurate classification of never-before-seen (i.e., zero-day) files using machine learning (ML). Consequently, efficacy of malware detectors is opaque, inhibiting end users from making informed decisions and researchers from targeting gaps in current detectors. In this article, we present a scientific evaluation of four prominent commercial malware detection tools to assist an organization with two primary questions: To what extent do ML-based tools accurately classify previously and never-before-seen files? Is purchasing a network-level malware detector worth the cost? To investigate, we tested each tool against 3,536 total files (2,554 or 72% malicious and 982 or 28% benign) of a variety of file types, including hundreds of malicious zero-days, polyglots, and APT-style files, delivered on multiple protocols. We present statistical results on detection time and accuracy, consider complementary analysis (using multiple tools together), and provide two novel applications of the recent cost–benefit evaluation procedure of Iannacone and Bridges. Although the ML-based tools are more effective at detecting zero-day files and executables, the signature-based tool might still be an overall better option. Both network-based tools provide substantial (simulated) savings when paired with either host tool, yet both show poor detection rates on protocols other than HTTP or SMTP. Our results show that all four tools have near-perfect precision but alarmingly low recall, especially on file types other than executables and office files: Thirty-seven percent of malware, including all polyglot files, were undetected. Priorities for researchers and takeaways for end users are given. Code for future use of the cost model is provided.

Learning Interpretable Microscopic Features of Tumor by Multi-task Adversarial CNNs To Improve Generalization

Adopting Convolutional Neural Networks (CNNs) in the daily routine of primary diagnosis requires not only near-perfect precision, but also a sufficient degree of generalization to data acquisition shifts and transparency. Existing CNN models act as black boxes, not ensuring to the physicians that important diagnostic features are used by the model. Building on top of successfully existing techniques such as multi-task learning, domain adversarial training and concept-based interpretability, this paper addresses the challenge of introducing diagnostic factors in the training objectives. Here we show that our architecture, by learning end-to-end an uncertainty-based weighting combination of multi-task and adversarial losses, is encouraged to focus on pathology features such as density and pleomorphism of nuclei, e.g. variations in size and appearance, while discarding misleading features such as staining differences. Our results on breast lymph node tissue show significantly improved generalization in the detection of tumorous tissue, with best average AUC 0.89 (0.01) against the baseline AUC 0.86 (0.005). By applying the interpretability technique of linearly probing intermediate representations, we also demonstrate that interpretable pathology features such as nuclei density are learned by the proposed CNN architecture, confirming the increased transparency of this model. This result is a starting point towards building interpretable multi-task architectures that are robust to data heterogeneity. Our code is available at &lt;a href='https://github.com/maragraziani/multitask_adversarial'&gt;https://github.com/maragraziani/multitask_adversarial&lt;/a&gt;

Deep feature selection using adaptive [formula omitted]-Hill Climbing aided whale optimization algorithm for lung and colon cancer detection

One of the most frightening and talked-about diseases in the modern world is cancer. Huge amounts of research are conducted worldwide to make this ailment less fearsome, be it by finding its cure, discovering ways to detect it in much earlier stages to reduce the mortality rate, or identifying precautions for humans to avoid it. The availability of large collections of biomedical and clinical data has ushered in the use of computer vision for cancer detection, especially for two of its most common types, lung and colon carcinomas. In this work, we present a framework wherein both deep learning and meta-heuristic approaches have been used for the prediction of colon or lung cancer, or both, from histopathological images with near-perfect precision. Initially, deep learning models, namely ResNet-18 for 2-class classification and EfficientNet-b4-widese for 3-class and 5-class classification, have been trained on the LC25000 dataset, followed by the extraction of deep features. The feature vector obtained from a deep learning model may have some redundancy. Hence, the selection of the most useful features has been done with the application of our proposed hybrid meta-heuristic optimization algorithm, AdBet-WOA (Whale optimization algorithm with integrated Adaptive β-Hill Climbing local search), utilizing which the Support Vector Machine (SVM) classifier classifies the colon cancer test data, lung cancer test data, and both combined with an accuracy of 99.99%, 99.97%, and 99.96%, respectively, matching the benchmark results comprehensively. For comparison, we have used a few independent as well as hybrid optimization algorithms. Our proposed approach succeeds greatly in reducing the number of features and also leads to better classification performance, as indicated by the obtained results. The relevant codes for our proposed approach are publicly available at: https://github.com/raj-1411/AdBet-WOA

Deterministic Modeling for Radiation Attenuation-integrated Radon Transform in Emission Computed Tomography: Algorithm, Curve Fitting Analysis, and Introduction of Attenuation Hadamard Matrix.

The purpose of the study is to propose an algorithm to implement and visualize radiation attenuation-integrated Radon transform based on Beer-Lambert law during emission computed tomography simulation using a deterministic model and also to perform image analysis on resulting images. Two types of phantoms are designed: plain-disk phantom and patterned-disk phantom. The large disk is filled with an activity of 5 units and the smaller disks have 10 units of activity of 99mTc isotope as an emission map. Three transmission maps for patterned-disk phantom are created with uniform linear attenuation coefficient. Phantoms are scanned with 360° and 180° acquisition arcs. Then, using the algorithm designed, the exponential Radon transform is implemented. After that, the projections are back-projected and filtered to generate tomographic slices. Finally, all slices are analyzed using profile plotting and curve fitting. Moreover, an attenuation Hadamard matrix is provided to facilitate attenuation modeling. The uniform intensity of activity in the phantoms is converted to a disk with progressively decreasing intensity from the periphery to the center in the tomographic slices. Similarly, the circles positioned more centrally appear less intense than those positioned in the periphery, despite all circles having equal activity. When the phantom is scanned in 180° arc, the circles closest to the camera are visualized more intensely. The profile curves of the slices generated by exponential Radon transformation are depicted as U-shaped in profile plotting and are fitted to a bi-exponential function with a near-perfect precision. The incorporation of radiation attenuation results in the development of more realistic models for quantification purposes.

Detecting the software usage on a compromised system: A triage solution for digital forensics

One of the challenges of digital forensics is the high volume of investigative cases. To address this problem, researchers have proposed various triage methods. Detecting the applications that have run on the compromised system under inspection can be an excellent triage method that gives the investigator an overview of the system. In this paper, we construct the signature of software usage on a system using file path artifacts. We propose a software signature detection engine (SSDE) to identify the usage of the software on the system under investigation. The SSDE consists of two subsystems: the signature construction subsystem, which builds the software signature using the TF-IDF weighting scheme, and the signature detection subsystem, which identifies the executed set of software on the target system. We consider several parameters with different values in the design of SSDEs, leading to more than 500 SSDE models. We test the SSDE models against 14 pseudo-real systems from the M57 Patents scenario and evaluate their performance. The experimental results show that about 38% of SSDE models achieve near-perfect Precision, and about 18% of them achieve near-perfect Recall. We introduce the top models and determine which parameter values lead to the superior models. Besides, we compare the SSDE models with some doc2vec-based models. The results show that SSDE models have higher average Precision, slightly lower average Recall, and much less computational time.

Fully Balanced SSFP Without an Endorectal Coil for Postimplant QA of MRI-Assisted Radiosurgery (MARS) of Prostate Cancer: A Prospective Study

To investigate fully balanced steady-state free precession (bSSFP) with optimized acquisition protocols for magnetic resonance imaging (MRI)-based postimplant quality assessment of low-dose-rate (LDR) prostate brachytherapy without an endorectal coil (ERC). Seventeen patients at a major academic cancer center who underwent MRI-assisted radiosurgery (MARS) LDR prostate cancer brachytherapy were imaged with moderate, high, or very high spatial resolution fully bSSFP MRIs without using an ERC. Between 1 and 3 signal averages (NEX) were acquired with acceleration factors (R) between 1 and 2, with the goal of keeping scan times between 4 and 6 minutes. Acquisitions with R >1 were reconstructed with parallel imaging and compressed sensing (PICS) algorithms. Radioactive seeds were identified by 3 medical dosimetrists. Additionally, some of the MRI techniques were implemented and tested at a community hospital; 3 patients underwent MARS LDR prostate brachytherapy and were imaged without an ERC. Increasing the in-plane spatial resolution mitigated partial volume artifacts and improved overall seed and seed marker visualization at the expense of reduced signal-to-noise ratio (SNR). The reduced SNR as a result of imaging at higher spatial resolution and without an ERC was partially compensated for by the multi-NEX acquisitions enabled by PICS. Resultant image quality was superior to the current clinical standard. All 3 dosimetrists achieved near-perfect precision and recall for seed identification in the 17 patients. The 3 postimplant MRIs acquired at the community hospital were sufficient to identify 208 out of 211 seeds implanted without reference to computed tomography (CT). Acquiring postimplant prostate brachytherapy MRI without an ERC has several advantages including better patient tolerance, lower costs, higher clinical throughput, and widespread access to precision LDR prostate brachytherapy. This prospective study confirms that the use of an ERC can be circumvented with fully bSSFP and advanced MRI scan techniques in a major academic cancer center and community hospital, potentially enabling postimplant assessment of MARS LDR prostate brachytherapy without CT.

Coupled whole-body rhythmic entrainment between two chimpanzees

Dance is an icon of human expression. Despite astounding diversity around the world’s cultures and dazzling abundance of reminiscent animal systems, the evolution of dance in the human clade remains obscure. Dance requires individuals to interactively synchronize their whole-body tempo to their partner’s, with near-perfect precision. This capacity is motorically-heavy, engaging multiple neural circuitries, but also dependent on an acute socio-emotional bond between partners. Hitherto, these factors helped explain why no dance forms were present amongst nonhuman primates. Critically, evidence for conjoined full-body rhythmic entrainment in great apes that could help reconstruct possible proto-stages of human dance is still lacking. Here, we report an endogenously-effected case of ritualized dance-like behaviour between two captive chimpanzees – synchronized bipedalism. We submitted video recordings to rigorous time-series analysis and circular statistics. We found that individual step tempo was within the genus’ range of “solo” bipedalism. Between-individual analyses, however, revealed that synchronisation between individuals was non-random, predictable, phase concordant, maintained with instantaneous centi-second precision and jointly regulated, with individuals also taking turns as “pace-makers”. No function was apparent besides the behaviour’s putative positive social affiliation. Our analyses show a first case of spontaneous whole-body entrainment between two ape peers, thus providing tentative empirical evidence for phylogenies of human dance. Human proto-dance, we argue, may have been rooted in mechanisms of social cohesion among small groups that might have granted stress-releasing benefits via gait-synchrony and mutual-touch. An external sound/musical beat may have been initially uninvolved. We discuss dance evolution as driven by ecologically-, socially- and/or culturally-imposed “captivity”.

A New Tool for Breast Anthropometric Measurements: Presentation and Validation for Women and Men

IntroductionAnthropometric measurements of breasts are crucial for planning surgical procedures; however, there are no practical solutions for their quick, digital performance. The aim of the study was to present and validate a self-designed web application BreastIdea (BI) designed for indirect breast anthropometry.MethodsTen male and 10 female volunteers had their chests measured directly according to the routine clinical practice. Then their chests were photographed in non-standardised conditions. Corresponding measurements were performed using BI. Accuracy and both relative and absolute reliability of BI measurements were investigated.ResultsBreast assessments using BI yielded highly accurate results and presented near-perfect precision when compared to direct anthropometric measurements of the breast. Indirect anthropometry eliminates the necessity to trace the body’s curves, which usually introduces a bias to linear measurements.ConclusionBI web application is a reliable tool for indirect breast measurements in a clinical setting, providing accurate results regardless of chest pathology and photograph standardisation.Level of Evidence IVThis journal requires that authors assign a level of evidence to each article. For a full description of these Evidence-Based Medicine ratings, please refer to the Table of Contents or the online Instructions to Authors www.springer.com/00266.

Towards a Knowledge-Based Recommender System for Linking Electronic Patient Records With Continuing Medical Education Information at the Point of Care

Given the limits of human memory, clinicians have trouble recalling therapeutic recommendations, even when the clinician previously judged that the information relevant for the care of a specific patient. To tackle this problem, we present a knowledge-based recommender system prototype that links the electronic patient records to clinical information, previously delivered to the target physician and judged to be potentially beneficial. We developed this prototype within the context of RxTx, a Canadian continuing medical education program. We apply a constraint-based recommendation strategy as follows: (1) clinical experts (taggers) map a set of therapeutic recommendations (called Highlights) to a requirement statement built from the standard clinical codes and supplementary demographic information, when applicable; (2) a matching system identifies patient-Highlights recommendation pairs through requirement satisfaction; and (3) given a patient record being examined, the recommended Highlights can be retrieved online at the point of care. We tested this prototype using electronic medical records from the Canadian Primary Care Sentinel Surveillance Network and 87 therapeutic Highlights from the RxTx collection, evaluating the system’s performance against a gold standard consisting of a two-expert consolidated patient-Highlight matching set for 150 patient records. The requirements-based recommendation system exhibits very high precision (mode: 1.0, 89% of the time; average precision: 0.95) and moderate recall (mode 1.0, 48.7% of the time; average recall: 0.61). The near-perfect precision minimizes the possibility of generating alert fatigue in physicians using the system. We note that more than half of the false negative results from the information being available in the text of the electronic medical records, but unavailable as a clinical code. The near-perfect precision over the tested patient set suggests that the system has the potential to deliver high-quality recommendations of clinical information at the point of care while being easily integrated within a continuing medical education program and the clinician’s workflow.

ZooBP

Given a heterogeneous network, with nodes of different types - e.g., products, users and sellers from an online recommendation site like Amazon - and labels for a few nodes ('honest', 'suspicious', etc), can we find a closed formula for Belief Propagation (BP), exact or approximate? Can we say whether it will converge? BP, traditionally an inference algorithm for graphical models, exploits so-called "network effects" to perform graph classification tasks when labels for a subset of nodes are provided; and it has been successful in numerous settings like fraudulent entity detection in online retailers and classification in social networks. However, it does not have a closed-form nor does it provide convergence guarantees in general. We propose ZooBP, a method to perform fast BP on undirected heterogeneous graphs with provable convergence guarantees. ZooBP has the following advantages: (1) Generality : It works on heterogeneous graphs with multiple types of nodes and edges; (2) Closed-form solution: ZooBP gives a closed-form solution as well as convergence guarantees; (3) Scalability: ZooBP is linear on the graph size and is up to 600× faster than BP, running on graphs with 3.3 million edges in a few seconds. (4) Effectiveness: Applied on real data (a F lipkart e-commerce network with users, products and sellers), ZooBP identifies fraudulent users with a near-perfect precision of 92.3 % over the top 300 results.

GeoScope

The First Law of Geography states "Everything is related to everything else, but near things are more related than distant things". This spatial significance has implications in various applications, trend detection being one of them. In this paper we propose a new algorithmic tool, GeoScope , to detect geo-trends. GeoScope is a data streams solution that detects correlations between topics and locations in a sliding window, in addition to analyzing topics and locations independently. GeoScope offers theoretical guarantees for detecting all trending correlated pairs while requiring only sub-linear space and running time. We perform various human validation tasks to demonstrate the value of GeoScope. The results show that human judges prefer GeoScope to the best performing baseline solution 4:1 in terms of the geographical significance of the presented information. As the Twitter analysis demonstrates, GeoScope successfully filters out topics without geo-intent and detects various local interests such as emergency events, political demonstrations or cultural events. Experiments on Twitter show that GeoScope has perfect recall and near-perfect precision.

Locating and parsing bibliographic references in HTML medical articles

The set of references that typically appear toward the end of journal articles is sometimes, though not always, a field in bibliographic (citation) databases. But even if references do not constitute such a field, they can be useful as a preprocessing step in the automated extraction of other bibliographic data from articles, as well as in computer-assisted indexing of articles. Automation in data extraction and indexing to minimize human labor is key to the affordable creation and maintenance of large bibliographic databases. Extracting the components of references, such as author names, article title, journal name, publication date and other entities, is therefore a valuable and sometimes necessary task. This paper describes a two-step process using statistical machine learning algorithms, to first locate the references in HTML medical articles and then to parse them. Reference locating identifies the reference section in an article and then decomposes it into individual references. We formulate this step as a two-class classification problem based on text and geometric features. An evaluation conducted on 500 articles drawn from 100 medical journals achieves near-perfect precision and recall rates for locating references. Reference parsing identifies the components of each reference. For this second step, we implement and compare two algorithms. One relies on sequence statistics and trains a Conditional Random Field. The other focuses on local feature statistics and trains a Support Vector Machine to classify each individual word, followed by a search algorithm that systematically corrects low confidence labels if the label sequence violates a set of predefined rules. The overall performance of these two reference-parsing algorithms is about the same: above 99% accuracy at the word level, and over 97% accuracy at the chunk level.

Stubborn Reliance on Intuition and Subjectivity in Employee Selection

The focus of this article is on implicit beliefs that inhibit adoption of selection decision aids (e.g., paper-and-pencil tests, structured interviews, mechanical combination of predictors). Understanding these beliefs is just as important as understanding organizational constraints to the adoption of selection technologies and may be more useful for informing the design of successful interventions. One of these is the implicit belief that it is theoretically possible to achieve near-perfect precision in predicting performance on the job. That is, people have an inherent resistance to analytical approaches to selection because they fail to view selection as probabilistic and subject to error. Another is the implicit belief that prediction of human behavior is improved through experience. This myth of expertise results in an overreliance on intuition and a reluctance to undermine one’s own credibility by using a selection decision aid.

The bag-of-frames approach to audio pattern recognition: A sufficient model for urban soundscapes but not for polyphonic music

The "bag-of-frames" approach (BOF) to audio pattern recognition represents signals as the long-term statistical distribution of their local spectral features. This approach has proved nearly optimal for simulating the auditory perception of natural and human environments (or soundscapes), and is also the most predominent paradigm to extract high-level descriptions from music signals. However, recent studies show that, contrary to its application to soundscape signals, BOF only provides limited performance when applied to polyphonic music signals. This paper proposes to explicitly examine the difference between urban soundscapes and polyphonic music with respect to their modeling with the BOF approach. First, the application of the same measure of acoustic similarity on both soundscape and music data sets confirms that the BOF approach can model soundscapes to near-perfect precision, and exhibits none of the limitations observed in the music data set. Second, the modification of this measure by two custom homogeneity transforms reveals critical differences in the temporal and statistical structure of the typical frame distribution of each type of signal. Such differences may explain the uneven performance of BOF algorithms on soundscapes and music signals, and suggest that their human perception rely on cognitive processes of a different nature.

Perfect and Near-Perfect Adaptation in a Model of Bacterial Chemotaxis

The signaling apparatus mediating bacterial chemotaxis can adapt to a wide range of persistent external stimuli. In many cases, the bacterial activity returns to its prestimulus level exactly, and this perfect adaptability is robust against variations in various chemotaxis protein concentrations. We model the bacterial chemotaxis signaling pathway, from ligand binding to CheY phosphorylation. By solving the steady-state equations of the model analytically, we derive a full set of conditions for the system to achieve perfect adaptation. The conditions related to the phosphorylation part of the pathway are discovered for the first time, while other conditions are generalizations of the ones found in previous works. Sensitivity of the perfect adaptation is evaluated by perturbing these conditions. We find that, even in the absence of some of the perfect adaptation conditions, adaptation can be achieved with near-perfect precision as a result of the separation of scales in both chemotaxis protein concentrations and reaction rates, or specific properties of the receptor distribution in different methylation states. Since near-perfect adaptation can be found in much larger regions of the parameter space than that defined by the perfect adaptation conditions, their existence is essential to understand robustness in bacterial chemotaxis.

Automatic Recognition of Phonetic Patterns in Speech

The mechanical recognition of ten spoken words (the digits) with near-perfect precision has been demonstrated under laboratory conditions for a single talker after optimum circuit adjustment for his voice. Such word recognition points to the possibility of automatic voice writing and other voice-controlled operations. The device described is based on two principles—the recognition of sustainable patterns of power versus frequency in speech, and the recognition of the durations of such phonetic patterns in words. An alphabet of only ten patterns is proven adequate for limited word recognition, at least in the case of a single talker. The successful performance of the ten-word digit recognizer for voice dialing of telephone numbers was recorded on a motion picture film shown in the oral presentation. Data from the film are discussed in this paper. This paper describes a working model of a phonetic pattern recognizer built at Bell Telephone Laboratories based on a detailed power-frequency spectrum analysis. Tests, to be discussed, indicate that this device, which supersedes an earlier model, leads to almost perfect digit recognition when set for a particular voice. The device also offers outstanding possibilities in speech transmission. The former application will be described in this paper and the latter in a companion paper.