Machine learning algorithms to classify spinal muscular atrophy subtypes

Neuromuscular disorders in the omics era

Neuromuscular disorders encompass a spectrum of conditions characterized by primary lesions within the peripheral nervous system, which include the anterior horn cell, peripheral nerve, neuromuscular junction, and muscle. In pediatrics, most of these disorders are linked to genetic causes. Despite the considerable progress, the diagnosis of these disorders remains a challenging due to wide clinical presentation, disease heterogeneity and rarity. It is noteworthy that certain neuromuscular disorders, once deemed untreatable, can now be effectively managed through novel therapies. Biomarkers emerge as indispensable tools, serving as objective measures that not only refine diagnostic accuracy but also provide guidance for therapeutic decision-making and the ongoing monitoring of long-term outcomes. Herein a comprehensive review of biomarkers in neuromuscular disorders is provided. We highlight the role of omics-based technologies that further characterize neuromuscular pathophysiology as well as identify potential therapeutic targets to guide treatment strategies.

Electrical impedance myography combined with quantitative assessment techniques in paretic muscle of stroke survivors: Insights and challenges

Aging is a non-modifiable risk factor for stroke and the global burden of stroke is continuing to increase due to the aging society. Muscle dysfunction, common sequela of stroke, has long been of research interests. Therefore, how to accurately assess muscle function is particularly important. Electrical impedance myography (EIM) has proven to be feasible to assess muscle impairment in patients with stroke in terms of micro structures, such as muscle membrane integrity, extracellular and intracellular fluids. However, EIM alone is not sufficient to assess muscle function comprehensively given the complex contributors to paretic muscle after an insult. This article discusses the potential to combine EIM and other common quantitative methods as ways to improve the assessment of muscle function in stroke survivors. Clinically, these combined assessments provide not only a distinct advantage for greater accuracy of muscle assessment through cross-validation, but also the physiological explanation on muscle dysfunction at the micro level. Different combinations of assessments are discussed with insights for different purposes. The assessments of morphological, mechanical and contractile properties combined with EIM are focused since changes in muscle structures, tone and strength directly reflect the muscle function of stroke survivors. With advances in computational technology, finite element model and machine learning model that incorporate multi-modal evaluation parameters to enable the establishment of predictive or diagnostic model will be the next step forward to assess muscle function for individual with stroke.

Using machine learning algorithms to enhance the diagnostic performance of electrical impedance myography

INTRODUCTION/AIMS

We assessed the classification performance of machine learning (ML) using multifrequency electrical impedance myography (EIM) values to improve upon diagnostic outcomes as compared to those based on a single EIM value.

METHODS

EIM data was obtained from unilateral excised gastrocnemius in eighty diseased mice (26 D2-mdx, Duchenne muscular dystrophy model, 39 SOD1G93A ALS model, and 15 db/db, a model of obesity-induced muscle atrophy) and 33 wild-type (WT) animals. We assessed the classification performance of a ML random forest algorithm incorporating all the data (multifrequency resistance, reactance and phase values) comparing it to the 50 kHz phase value alone.

RESULTS

ML outperformed the 50 kHz analysis as based on receiver-operating characteristic curves and measurement of the area under the curve (AUC). For example, comparing all diseases together versus WT from the test set outputs, the AUC was 0.52 for 50 kHz phase, but was 0.94 for the ML model. Similarly, when comparing ALS versus WT, the AUCs were 0.79 for 50 kHz phase and 0.99 for ML.

DISCUSSION

Multifrequency EIM utilizing ML improves upon classification compared to that achieved with a single-frequency value. ML approaches should be considered in all future basic and clinical diagnostic applications of EIM.

Tensor electrical impedance myography identifies clinically relevant features in amyotrophic lateral sclerosis

Objective.Electrical impedance myography (EIM) shows promise as an effective biomarker in amyotrophic lateral sclerosis (ALS). EIM applies multiple input frequencies to characterise muscle properties, often via multiple electrode configurations. Herein, we assess if non-negative tensor factorisation (NTF) can provide a framework for identifying clinically relevant features within a high dimensional EIM dataset.Approach.EIM data were recorded from the tongue of healthy and ALS diseased individuals. Resistivity and reactivity measurements were made for 14 frequencies, in three electrode configurations. This gives 84 (2 × 14 × 3) distinct data points per participant. NTF was applied to the dataset for dimensionality reduction, termed tensor EIM. Significance tests, symptom correlation and classification approaches were explored to compare NTF to using all raw data and feature selection.Main Results.Tensor EIM provides highly significant differentiation between healthy and ALS patients (p< 0.001, AUROC = 0.78). Similarly tensor EIM differentiates between mild and severe disease states (p< 0.001, AUROC = 0.75) and significantly correlates with symptoms (ρ= 0.7,p< 0.001). A trend of centre frequency shifting to the right was identified in diseased spectra, which is in line with the electrical changes expected following muscle atrophy.Significance.Tensor EIM provides clinically relevant metrics for identifying ALS-related muscle disease. This procedure has the advantage of using the whole spectral dataset, with reduced risk of overfitting. The process identifies spectral shapes specific to disease allowing for a deeper clinical interpretation.

A Machine-Learning Method of Predicting Vital Capacity Plateau Value for Ventilatory Pump Failure Based on Data Mining

Ventilatory pump failure is a common cause of death for patients with neuromuscular diseases. The vital capacity plateau value (VCPLAT) is an important indicator to judge the status of ventilatory pump failure for patients with congenital myopathy, Duchenne muscular dystrophy and spinal muscular atrophy. Due to the complex relationship between VCPLAT and the patient's own condition, it is difficult to predict the VCPLAT for pediatric disease from a medical perspective. We established a VCPLAT prediction model based on data mining and machine learning. We first performed the correlation analysis and recursive feature elimination with cross-validation (RFECV) to provide high-quality feature combinations. Based on this, the Light Gradient Boosting Machine (LightGBM) algorithm was to establish a prediction model with powerful performance. Finally, we verified the validity and superiority of the proposed method via comparison with other prediction models in similar works. After 10-fold cross-validation, the proposed prediction method had the best performance and its explained variance score (EVS), mean absolute error (MAE), mean squared error (MSE), root mean square error (RMSE), median absolute error (MedAE) and R2 were 0.949, 0.028, 0.002, 0.045, 0.015 and 0.948, respectively. It also performed well on test datasets. Therefore, it can accurately and effectively predict the VCPLAT, thereby determining the severity of the condition to provide auxiliary decision-making for doctors in clinical diagnosis and treatment.

A Machine Learning Approach to the Interpretation of Cardiopulmonary Exercise Tests: Development and Validation

Objective

At present, there is no consensus on the best strategy for interpreting the cardiopulmonary exercise test's (CPET) results. This study is aimed at assessing the potential of using computer-aided algorithms to evaluate CPET data for identifying chronic heart failure (CHF) and chronic obstructive pulmonary disease (COPD).

Methods

Data from 234 CPET files from the Pulmonary Institute, at Sheba Medical Center, and the Givat-Washington College, both in Israel, were selected for this study. The selected CPET files included patients with confirmed primary CHF (n = 73), COPD (n = 75), and healthy subjects (n = 86). Of the 234 CPETs, 150 (50 in each group) tests were used for the support vector machine (SVM) learning stage, and the remaining 84 tests were used for the model validation. The performance of the SVM interpretive module was assessed by comparing its interpretation output with the conventional clinical diagnosis using distribution analysis.

Results

The disease classification results show that the overall predictive power of the proposed interpretive model ranged from 96% to 100%, indicating very high predictive power. Furthermore, the sensitivity, specificity, and overall precision of the proposed interpretive module were 99%, 99%, and 99%, respectively.

Conclusions

The proposed new computer-aided CPET interpretive module was found to be highly sensitive and specific in classifying patients with CHF or COPD, or healthy. Comparable modules may well be applied to additional and larger populations (pathologies and exercise limitations), thereby making this tool powerful and clinically applicable.

Multi-dimensional electrical impedance myography of the tongue as a potential biomarker for amyotrophic lateral sclerosis

OBJECTIVE

In amyotrophic lateral sclerosis (ALS) bulbar disease biomarkers are lacking. We evaluated a novel tongue electrical impedance myography (EIM) system, utilising both 2D and 3D electrode configurations for detection of tongue pathology.

METHODS

Longitudinal multi-frequency phase angle spectra were recorded from 41 patients with ALS (baseline, 3 and 6 months) and 30 healthy volunteers (baseline and 6 months). ALS functional rating scale-revised (ALSFRS-R) data and quantitative tongue strength measurements were collected. EIM data were analysed for reliability (intra-class correlation coefficient; ICC) and differences between patients and volunteers ascertained using both univariate (Mann-Whitney U test) and multivariate techniques (feature selection and L2 norm).

RESULTS

The device produced highly reliable data (pooled ICC: 0.836). Significant EIM differences were apparent between ALS patients and healthy volunteers (P < 0.001). EIM data demonstrated a significant relationship to tongue strength and bulbar ALSFRS-R scores (P < 0.015). The EIM recordings revealed a group level longitudinal change over 6 months and consistently identified patients in whom symptoms or tongue strength changed.

CONCLUSIONS

The novel EIM tongue system produces reliable data and can differentiate between healthy muscle and ALS-related disease.

SIGNIFICANCE

Tongue EIM utilising multiple frequencies and electrode configurations has potential as a bulbar disease biomarker in ALS.

Machine Learning in Amyotrophic Lateral Sclerosis: Achievements, Pitfalls, and Future Directions

Background: Amyotrophic Lateral Sclerosis (ALS) is a relentlessly progressive neurodegenerative condition with limited therapeutic options at present. Survival from symptom onset ranges from 3 to 5 years depending on genetic, demographic, and phenotypic factors. Despite tireless research efforts, the core etiology of the disease remains elusive and drug development efforts are confounded by the lack of accurate monitoring markers. Disease heterogeneity, late-stage recruitment into pharmaceutical trials, and inclusion of phenotypically admixed patient cohorts are some of the key barriers to successful clinical trials. Machine Learning (ML) models and large international data sets offer unprecedented opportunities to appraise candidate diagnostic, monitoring, and prognostic markers. Accurate patient stratification into well-defined prognostic categories is another aspiration of emerging classification and staging systems. Methods: The objective of this paper is the comprehensive, systematic, and critical review of ML initiatives in ALS to date and their potential in research, clinical, and pharmacological applications. The focus of this review is to provide a dual, clinical-mathematical perspective on recent advances and future directions of the field. Another objective of the paper is the frank discussion of the pitfalls and drawbacks of specific models, highlighting the shortcomings of existing studies and to provide methodological recommendations for future study designs. Results: Despite considerable sample size limitations, ML techniques have already been successfully applied to ALS data sets and a number of promising diagnosis models have been proposed. Prognostic models have been tested using core clinical variables, biological, and neuroimaging data. These models also offer patient stratification opportunities for future clinical trials. Despite the enormous potential of ML in ALS research, statistical assumptions are often violated, the choice of specific statistical models is seldom justified, and the constraints of ML models are rarely enunciated. Conclusions: From a mathematical perspective, the main barrier to the development of validated diagnostic, prognostic, and monitoring indicators stem from limited sample sizes. The combination of multiple clinical, biofluid, and imaging biomarkers is likely to increase the accuracy of mathematical modeling and contribute to optimized clinical trial designs.

Machine learning for medical ultrasound: status, methods, and future opportunities

Ultrasound (US) imaging is the most commonly performed cross-sectional diagnostic imaging modality in the practice of medicine. It is low-cost, non-ionizing, portable, and capable of real-time image acquisition and display. US is a rapidly evolving technology with significant challenges and opportunities. Challenges include high inter- and intra-operator variability and limited image quality control. Tremendous opportunities have arisen in the last decade as a result of exponential growth in available computational power coupled with progressive miniaturization of US devices. As US devices become smaller, enhanced computational capability can contribute significantly to decreasing variability through advanced image processing. In this paper, we review leading machine learning (ML) approaches and research directions in US, with an emphasis on recent ML advances. We also present our outlook on future opportunities for ML techniques to further improve clinical workflow and US-based disease diagnosis and characterization.

Automated diagnosis of myositis from muscle ultrasound: Exploring the use of machine learning and deep learning methods

OBJECTIVE

To evaluate the use of ultrasound coupled with machine learning (ML) and deep learning (DL) techniques for automated or semi-automated classification of myositis.

METHODS

Eighty subjects comprised of 19 with inclusion body myositis (IBM), 14 with polymyositis (PM), 14 with dermatomyositis (DM), and 33 normal (N) subjects were included in this study, where 3214 muscle ultrasound images of 7 muscles (observed bilaterally) were acquired. We considered three problems of classification including (A) normal vs. affected (DM, PM, IBM); (B) normal vs. IBM patients; and (C) IBM vs. other types of myositis (DM or PM). We studied the use of an automated DL method using deep convolutional neural networks (DL-DCNNs) for diagnostic classification and compared it with a semi-automated conventional ML method based on random forests (ML-RF) and "engineered" features. We used the known clinical diagnosis as the gold standard for evaluating performance of muscle classification.

RESULTS

The performance of the DL-DCNN method resulted in accuracies ± standard deviation of 76.2% ± 3.1% for problem (A), 86.6% ± 2.4% for (B) and 74.8% ± 3.9% for (C), while the ML-RF method led to accuracies of 72.3% ± 3.3% for problem (A), 84.3% ± 2.3% for (B) and 68.9% ± 2.5% for (C).

CONCLUSIONS

This study demonstrates the application of machine learning methods for automatically or semi-automatically classifying inflammatory muscle disease using muscle ultrasound. Compared to the conventional random forest machine learning method used here, which has the drawback of requiring manual delineation of muscle/fat boundaries, DCNN-based classification by and large improved the accuracies in all classification problems while providing a fully automated approach to classification.

Intraoperative Diagnosis Support Tool for Serous Ovarian Tumors Based on Microarray Data Using Multicategory Machine Learning

OBJECTIVES

Serous borderline ovarian tumors (SBOTs) are a subtype of serous ovarian carcinoma with atypical proliferation. Frozen-section diagnosis has been used as an intraoperative diagnosis tool in supporting the fertility-sparing surgery by diagnosing SBOTs with accuracy of 48% to 79%. Using DNA microarray technology, we designed multicategory classification models to support frozen-section diagnosis within 30 minutes.

MATERIALS AND METHODS

We systematically evaluated 6 machine learning algorithms and 3 feature selection methods using 5-fold cross-validation and a grid search on microarray data obtained from the National Center for Biotechnology Information. To validate the models and selected biomarkers, expression profiles were analyzed in tissue samples obtained from the Yonsei University College of Medicine.

RESULTS

The best accuracy of the optimal machine learning model was 97.3%. In addition, 5 features, including the expression of the putative biomarkers SNTN and AOX1, were selected to differentiate between normal, SBOT, and serous ovarian carcinoma groups. Different expression levels of SNTN and AOX1 were validated by real-time quantitative reverse-transcription polymerase chain reaction, Western blotting, and immunohistochemistry. A multinomial logistic regression model using SNTN and AOX1 alone was used to construct a simple-to-use equation that gave a diagnostic test accuracy of 91.9%.

CONCLUSIONS

We identified 2 biomarkers, SNTN and AOX1, that are likely involved in the pathogenesis and progression of ovarian tumors. An accurate diagnosis of ovarian tumor subclasses by application of the equation in conjunction with expression analysis of SNTN and AOX1 would offer a new accurate diagnosis tool in conjunction with frozen-section diagnosis within 30 minutes.

The neuromuscular impact of symptomatic SMN restoration in a mouse model of spinal muscular atrophy

BACKGROUND

Significant advances in the development of SMN-restoring therapeutics have occurred since 2010 when very effective biological treatments were reported in mouse models of spinal muscular atrophy. As these treatments are applied in human clinical trials, there is pressing need to define quantitative assessments of disease progression, treatment stratification, and therapeutic efficacy. The electrophysiological measures Compound Muscle Action Potential and Motor Unit Number Estimation are reliable measures of nerve function. In both the SMN∆7 mouse and a pig model of spinal muscular atrophy, early SMN restoration results in preservation of electrophysiological measures. Currently, clinical trials are underway in patients at post-symptomatic stages of disease progression. In this study, we present results from both early and delayed SMN restoration using clinically-relevant measures including electrical impedance myography, compound muscle action potential, and motor unit number estimation to quantify the efficacy and time-sensitivity of SMN-restoring therapy.

METHODS

SMA∆7 mice were treated via intracerebroventricular injection with antisense oligonucleotides targeting ISS-N1 to increase SMN protein from the SMN2 gene on postnatal day 2, 4, or 6 and compared with sham-treated spinal muscular atrophy and control mice. Compound muscle action potential and motor unit number estimation of the triceps surae muscles were performed at day 12, 21, and 30 by a single evaluator blinded to genotype and treatment. Similarly, electrical impedance myography was measured on the biceps femoris muscle at 12days for comparison.

RESULTS

Electrophysiological measures and electrical impedance myography detected significant differences at 12days between control and late-treated (4 or 6days) and sham-treated spinal muscular atrophy mice, but not in mice treated at 2days (p<0.01). EIM findings paralleled and correlated with compound muscle action potential and motor unit number estimation (r=0.61 and r=0.50, respectively, p<0.01). Longitudinal measures at 21 and 30days show that symptomatic therapy results in reduced motor unit number estimation associated with delayed normalization of compound muscle action potential.

CONCLUSIONS

The incomplete effect of symptomatic treatment is accurately identified by both electrophysiological measures and electrical impedance myography. There is strong correlation between these measures and with weight and righting reflex. This study predicts that measures of compound muscle action potential, motor unit number estimation, and electrical impedance myography are promising biomarkers of treatment stratification and effect for future spinal muscular atrophy trials. The ease of application and simplicity of electrical impedance myography compared with standard electrophysiological measures may be particularly valuable in future pediatric clinical trials.

Multicategory classification of 11 neuromuscular diseases based on microarray data using support vector machine

We applied multicategory machine learning methods to classify 11 neuromuscular disease groups and one control group based on microarray data. To develop multicategory classification models with optimal parameters and features, we performed a systematic evaluation of three machine learning algorithms and four feature selection methods using three-fold cross validation and a grid search. This study included 114 subjects of 11 neuromuscular diseases and 31 subjects of a control group using microarray data with 22,283 probe sets from the National Center for Biotechnology Information (NCBI). We obtained an accuracy of 100%, relative classifier information (RCI) of 1.0, and a kappa index of 1.0 by applying the models of support vector machines one-versus-one (SVM-OVO), SVM one-versus-rest (OVR), and directed acyclic graph SVM (DAGSVM), using the ratio of genes between categories to within-category sums of squares (BW) feature selection method. Each of these three models selected only four features to categorize the 12 groups, resulting in a time-saving and cost-effective strategy for diagnosing neuromuscular diseases. In addition, a gene symbol, SPP1 was selected as the top-ranked gene by the BW method. We confirmed relationships between the gene (SPP1) and Duchenne muscular dystrophy (DMD) from a previous study. With our models as clinically helpful tools, neuromuscular diseases could be classified quickly using a computer, thereby giving a time-saving, cost-effective, and accurate diagnosis.

Quantitative muscle ultrasound measures rapid declines over time in children with SMA type 1

Muscles are small in spinal muscular atrophy (SMA). It is not known if muscle size changes over time in SMA type 1. We quantified changes over time in muscle size and echointensity during two repeated ultrasound examinations of unilateral proximal (biceps brachii/brachialis and quadriceps) and distal (anterior forearm flexors and tibialis anterior) muscles in three children with SMA type 1. We compared muscle thickness (MT) to body weight-dependent normal reference values. Children were 1, 6, and 11months old at baseline and had 2, 2 and 4 months between ultrasound examinations, respectively. At baseline, MT was normal for weight in all muscles except an atrophic quadriceps in the oldest child. MT decreased and echointensity increased (worsened) over time. At follow up, MT was below normal for weight in the quadriceps in all three children, in the biceps/brachioradialis in two, and in the anterior forearm in one. Tibialis anterior MT remained normal for weight in all three children. Muscle echointensity increased over time in all muscles and, on average, more than doubled in two children. In children with SMA type 1, muscle atrophies and becomes hyperechoic over time. Quantitative muscle ultrasound measures disease progression in SMA type 1 that warrants additional study in more children.

Composite biomarkers for assessing Duchenne muscular dystrophy: an initial assessment

BACKGROUND

Compared with individual parameters, composite biomarkers may provide a more effective means for monitoring disease progression and the effects of therapy in clinical trials than single measures. In this study, we built composite biomarkers for use in Duchenne muscular dystrophy by combining values from two objective measures of disease severity: electrical impedance myography and quantitative ultrasound and evaluating how well they correlated to standard functional measures.

METHODS

Using data from an ongoing study of electrical impedance myography and quantitative ultrasound in 31 Duchenne muscular dystrophy and 26 healthy boys aged 2-14 years, we combined data sets by first creating z scores based on the normal subject data and then using simple mathematical operations (addition and multiplication) to create composite measures. These composite scores were then correlated to age and standard measures of function including the 6-minute walk test, the North Star Ambulatory Assessment, and handheld dynamometry.

RESULTS

Combining data sets resulted in stronger correlations with all four outcomes than for either electrical impedance myography or quantitative ultrasound alone in six of eight instances. These improvements reached statistical significance (P < 0.05) in several cases. For example, the correlation coefficient for the composite measure with the North Star Ambulatory Assessment was 0.79 but was only 0.66 and 0.67 (respectively) for gray scale level and electrical impedance myography separately.

CONCLUSIONS

Arithmetically derived composite scores can provide stronger correlations to functional measures than isolated biomarkers. Longitudinal study of such composite markers in Duchenne muscular dystrophy clinical trials is warranted.

Cross-sectional evaluation of electrical impedance myography and quantitative ultrasound for the assessment of Duchenne muscular dystrophy in a clinical trial setting

BACKGROUND

Electrical impedance myography and quantitative ultrasound are two noninvasive, painless, and effort-independent approaches for assessing neuromuscular disease. Both techniques have potential to serve as useful biomarkers in clinical trials in Duchenne muscular dystrophy. However, their comparative sensitivity to disease status and how they relate to one another are unknown.

METHODS

We performed a cross-sectional analysis of electrical impedance myography and quantitative ultrasound in 24 healthy boys and 24 with Duchenne muscular dystrophy, aged 2 to 14 years with trained research assistants performing all measurements. Three upper and three lower extremity muscles were studied unilaterally in each child, and the data averaged for each individual.

RESULTS

Both electrical impedance myography and quantitative ultrasound differentiated healthy boys from those with Duchenne muscular dystrophy (P < 0.001 for both). Quantitative ultrasound values correlated with age in Duchenne muscular dystrophy boys (rho = 0.45; P = 0.029), whereas electrical impedance myography did not (rho = -0.31; P = 0.14). However, electrical impedance myography phase correlated with age in healthy boys (rho = 0.51; P = 0.012), whereas quantitative ultrasound did not (rho = -0.021; P = 0.92). In Duchenne muscular dystrophy boys, electrical impedance myography phase correlated with the North Star Ambulatory Assessment (rho = 0.65; P = 0.022); quantitative ultrasound revealed a near-significant association (rho = -0.56; P = 0.060). The two technologies trended toward a moderate correlation with one another in the Duchenne muscular dystrophy cohort but not in the healthy group (rho = -0.40; P = 0.054 and rho = -0.32; P = 0.13, respectively).

CONCLUSIONS

Electrical impedance myography and quantitative ultrasound are complementary modalities for the assessment of boys with Duchenne muscular dystrophy; further study and application of these two modalities alone or in combination in a longitudinal fashion are warranted.

SMA-MAP: a plasma protein panel for spinal muscular atrophy

Objectives

Spinal Muscular Atrophy (SMA) presents challenges in (i) monitoring disease activity and predicting progression, (ii) designing trials that allow rapid assessment of candidate therapies, and (iii) understanding molecular causes and consequences of the disease. Validated biomarkers of SMA motor and non-motor function would offer utility in addressing these challenges. Our objectives were (i) to discover additional markers from the Biomarkers for SMA (BforSMA) study using an immunoassay platform, and (ii) to validate the putative biomarkers in an independent cohort of SMA patients collected from a multi-site natural history study (NHS).

Methods

BforSMA study plasma samples (N = 129) were analyzed by immunoassay to identify new analytes correlating to SMA motor function. These immunoassays included the strongest candidate biomarkers identified previously by chromatography. We selected 35 biomarkers to validate in an independent cohort SMA type 1, 2, and 3 samples (N = 158) from an SMA NHS. The putative biomarkers were tested for association to multiple motor scales and to pulmonary function, neurophysiology, strength, and quality of life measures. We implemented a Tobit model to predict SMA motor function scores.

Results

12 of the 35 putative SMA biomarkers were significantly associated (p<0.05) with motor function, with a 13^th analyte being nearly significant. Several other analytes associated with non-motor SMA outcome measures. From these 35 biomarkers, 27 analytes were selected for inclusion in a commercial panel (SMA-MAP) for association with motor and other functional measures.

Conclusions

Discovery and validation using independent cohorts yielded a set of SMA biomarkers significantly associated with motor function and other measures of SMA disease activity. A commercial SMA-MAP biomarker panel was generated for further testing in other SMA collections and interventional trials. Future work includes evaluating the panel in other neuromuscular diseases, for pharmacodynamic responsiveness to experimental SMA therapies, and for predicting functional changes over time in SMA patients.