Disease Delineation for Multiple Sclerosis, Friedreich Ataxia, and Healthy Controls Using Supervised Machine Learning on Speech Acoustics

Neurodegenerative disease often affects speech. Speech acoustics can be used as objective clinical markers of pathology. Previous investigations of pathological speech have primarily compared controls with one specific condition and excluded comorbidities. We broaden the utility of speech markers by examining how multiple acoustic features can delineate diseases. We used supervised machine learning with gradient boosting (CatBoost) to delineate healthy speech from speech of people with multiple sclerosis or Friedreich ataxia. Participants performed a diadochokinetic task where they repeated alternating syllables. We subjected 74 spectral and temporal prosodic features from the speech recordings to machine learning. Results showed that Friedreich ataxia, multiple sclerosis and healthy controls were all identified with high accuracy (over 82%). Twenty-one acoustic features were strong markers of neurodegenerative diseases, falling under the categories of spectral qualia, spectral power, and speech rate. We demonstrated that speech markers can delineate neurodegenerative diseases and distinguish healthy speech from pathological speech with high accuracy. Findings emphasize the importance of examining speech outcomes when assessing indicators of neurodegenerative disease. We propose large-scale initiatives to broaden the scope for differentiating other neurological diseases and affective disorders.

The cost and impact of distributing naloxone to people who are prescribed opioids to prevent opioid-related deaths: findings from a modelling study

BACKGROUND AND AIMS

Although most opioid-related mortality in Australia involves prescription opioids, most research to understand the impact of naloxone supply on opioid-related mortality has focused upon people who inject heroin. We aimed to examine the cost and probable impact of up-scaling naloxone supply to people who are prescribed opioids.

DESIGN

Decision-tree model. Four scenarios were compared with a baseline scenario (the current status quo): naloxone scale-up between 2020 and 2030 to reach 30 or 90% coverage by 2030, among the subgroups of people prescribed either ≥ 50 or ≥ 100 mg of oral morphine equivalents (OME).

SETTING

Australia.

PARTICIPANTS

People who are prescribed opioids.

MEASUREMENTS

Possible deaths averted, costs (ambulance and naloxone distribution) and cost per life saved for different scenarios of naloxone scale-up.

FINDINGS

Maintaining the status quo, there would be an estimated 7478 [uncertainty interval (UI) = 6868-8275] prescription opioid overdose deaths between 2020 and 2030, resulting in Australian dollars (A$)51.9 million (49.4, 56.0) in ambulance costs. If naloxone were scaled-up to 90% of people prescribed > 50 mg OME, an estimated 657 (UI = 245, 1489) deaths could be averted between 2020 and 2030 (a 20% reduction in the final year of the model compared with the no naloxone scenario), with a cost of A$43 600 (20 800-110 500) per life saved. If naloxone were scaled-up to 30% of people prescribed > 50 mg OME an estimated 219 (82-496) deaths could be averted with the same cost per live saved. If naloxone were restricted to those prescribed > 100 mg OME, an estimated 130 (UI = 44-289) deaths would be averted if scaled-up to 30% or 390 (UI = 131-866) deaths averted if scaled-up to 90%, with the cost per life saved for both scenarios A$38 200 (UI = 12 400-97 400).

CONCLUSION

In Australia, scaling-up take-home naloxone by 2030 to reach 90% of people prescribed daily doses of ≥ 50 mg of oral morphine equivalents would be cost-effective and save more than 650 lives.

The Melbourne Safe Injecting Room Attracted People Most in Need of Its Service

INTRODUCTION

In 2018, the first Medically Supervised Injecting Room in Melbourne, Australia was officially opened. This study assessed whether this facility attracted people who inject drugs, who were socially vulnerable, and who engaged in drug-related behaviors associated with increased morbidity and mortality risk.

METHODS

This was a cross-sectional analysis of the frequency of Medically Supervised Injecting Room use during the first 18 months after opening (July 2018-December 2019) among 658 people who inject drugs participating in the Melbourne Injecting Drug User Cohort Study (SuperMIX). To examine the differences between no Medically Supervised Injecting Room use, infrequent use (<50% injections within the facility), and frequent use (≥50% of injections within the facility), RRRs were estimated using bivariate multinomial logistic regression analyses and postestimation Wald tests. Analyses were conducted in 2020.

RESULTS

A total of 451 participants (68%) reported no Medically Supervised Injecting Room use, 142 (22%) reported infrequent use, and 65 (10%) reported frequent use. Participants who reported either infrequent or frequent use of the facility were more socially vulnerable (e.g., more often homeless) and more likely to report risky drug-related behaviors and poor health outcomes than those who reported no use. Participants who reported frequent use of the facility were also more likely to live close to the facility than those reporting infrequent use.

CONCLUSIONS

The Melbourne Medically Supervised Injecting Room attracted socially marginalized people who inject drugs who are most at risk of harms related to injecting drug use and therefore who are most in need of the service. To determine the long-term impact use of this facility on key health outcomes such as overdose, future studies should consider the differences in vulnerability and risk behavior of people who inject drugs who use the Medically Supervised Injecting Room when examining the outcomes associated with the use of the facility.

A virtual experiment improved students' understanding of physiological experimental processes ahead of a live inquiry-based practical class

Physiology is commonly taught through direct experience and observation of scientific phenomena in "hands-on" practical laboratory classes. The value of such classes is limited by students' lack of understanding of the underlying theoretical concepts and their lack of confidence with the experimental techniques. In our experience, students follow experimental steps as if following a recipe, without giving thought to the underlying theory and the relationship between the experimental procedure and the research hypotheses. To address this issue, and to enhance student learning, we developed an online virtual experiment for students to complete before an inquiry-based practical. The virtual experiment and "live" practical laboratory were an investigation of how autonomic nerves control contractions in the isolated rabbit ileum. We hypothesized that the virtual experiment would support students' understanding of the physiological concepts, as well as the experimental design associated with the practical. Anonymous survey data and usage analytics showed that most students engaged with the virtual experiment. Students thought that it helped them to understand the practical physiological concepts and experimental design, with self-reported time spent on the virtual experiment (and not on lectures or practical class notes) a significant predictor of their understanding. This novel finding provides evidence that virtual experiments can contribute to students' research skills development. Our results indicate that self-paced online virtual experiments are an effective way to enhance student understanding of physiological concepts and experimental processes, allowing for a more realistic experience of the scientific method and a more effective use of time in practical classes.

Short-Term Attractive Tilt Aftereffects Predicted by a Recurrent Network Model of Primary Visual Cortex

Adaptation is a multi-faceted phenomenon that is of interest in terms of both its function and its potential to reveal underlying neural processing. Many behavioral studies have shown that after exposure to an oriented adapter the perceived orientation of a subsequent test is repulsed away from the orientation of the adapter. This is the well-known Tilt Aftereffect (TAE). Recently, we showed that the dynamics of recurrently connected networks may contribute substantially to the neural changes induced by adaptation, especially on short time scales. Here we extended the network model and made the novel behavioral prediction that the TAE should be attractive, not repulsive, on a time scale of a few 100 ms. Our experiments, using a novel adaptation protocol that specifically targeted adaptation on a short time scale, confirmed this prediction. These results support our hypothesis that recurrent network dynamics may contribute to short-term adaptation. More broadly, they show that understanding the neural processing of visual inputs that change on the time scale of a typical fixation requires a detailed analysis of not only the intrinsic properties of neurons, but also the slow and complex dynamics that emerge from their recurrent connectivity. We argue that this is but one example of how even simple recurrent networks can underlie surprisingly complex information processing, and are involved in rudimentary forms of memory, spatio-temporal integration, and signal amplification.

Adaptation without Plasticity

Sensory adaptation is a phenomenon in which neurons are affected not only by their immediate input but also by the sequence of preceding inputs. In visual cortex, for example, neurons shift their preferred orientation after exposure to an oriented stimulus. This adaptation is traditionally attributed to plasticity. We show that a recurrent network generates tuning curve shifts observed in cat and macaque visual cortex, even when all synaptic weights and intrinsic properties in the model are fixed. This demonstrates that, in a recurrent network, adaptation on timescales of hundreds of milliseconds does not require plasticity. Given the ubiquity of recurrent connections, this phenomenon likely contributes to responses observed across cortex and shows that plasticity cannot be inferred solely from changes in tuning on these timescales. More broadly, our findings show that recurrent connections can endow a network with a powerful mechanism to store and integrate recent contextual information.