By the Numbers: Machine studying, dementia hyperlink, antipsychotics whereas pregnant | Spectrum
Illustration by Laurène Boglio
Welcome to this month’s edition of the By the Numbers newsletter. At Spectrum, we do our best to summarize the latest in autism research – and sometimes the best summary comes in the form of a chart or map. In this newsletter we put interesting new research results in a nutshell, which are conveyed most concisely through data visualizations.
Let us know what you think of the newsletter at niko@spectrumnews.org or tell us about your own data-rich work.
Machine learning and neuroimaging
Machine learning models may be better at detecting signs of autism in brain scans when the data sets include women, according to a new analysis of data from 44 studies.
“We want to improve the data used for machine learning because, you know, garbage in, garbage out,” says senior investigator Sotirios Bisdas, a neuroradiologist at the National Hospital for Neurology in London, England. According to Bisdas, the aim of the new study was “to understand whether there are systemic biases that influence the quality of the data and can subsequently influence the results”.
It turned out that the answer is yes: some types of neuroimaging data, like electroencephalography, are potentially better than others for training machine learning models to detect autism, the study shows. Involving women and data from multiple imaging technologies can also improve the sensitivity and specificity of some machine learning algorithms.
It’s unlikely that machine learning will help diagnose autism anytime soon. There are huge differences in the way autism shows up in boys and girls, how the condition is diagnosed, how neuroimaging is done, and what machine learning algorithms are used to draw conclusions from data.
In order to improve the accuracy of machine learning algorithms, researchers should first standardize imaging processes, says Bisdas.
“It doesn’t make sense to wait for more image data if we don’t standardize your collection,” says Bisdas. “Yes, we need more data, but we need more quality data.”
The results appeared in Neuroradiology in August.
Early onset dementia and autism
According to a new study, around 4 percent of adults with autism were diagnosed with early-onset dementia between 2008 and 2012, compared with less than 1 percent of non-autistic adults. The prevalence is more than 5 percent among people with autism and intellectual disabilities.
The data is from Medicaid Analytic eXtract and included more than 1.2 million adults who received Medicaid benefits between the ages of 30 and 64 years. More than 12,000 people in the dataset have autism, of whom 511 were diagnosed with dementia before the age of 65; those with autism had a higher prevalence of early-onset dementia in blacks than in whites; it was also higher for those who lived in rural areas compared to urban dwellers.
The results appeared in Autism Research in August.
Children of women who take antipsychotics such as aripiprazole or risperidone during pregnancy have, according to one in JAMA. published study neither an increased risk of autism or attention deficit hyperactivity disorder nor are they more likely to be born prematurely or underweight Internal Medicine in August.
Prescribing antipsychotics during pregnancy can help prevent potentially dangerous psychotic episodes and ensure an expectant mother can fend for herself, says Mady Hornig, an adjunct professor of epidemiology at Columbia University who was not involved in the study . “We certainly don’t want to be rash about taking medication during pregnancy, but we also want to weigh the effects of not treating it.”
Any apparent effects of antipsychotics on a developing fetus are likely due to the condition being treated rather than the treatment itself, the study shows. That’s because doctors prescribe medication for a reason, which can create additional differences between women taking and not taking medication and skew study results. This explains why women who received medication during pregnancy appear to be at increased risk of having children with autism, as shown in the table above.
The results were published in JAMA Internal Medicine in August.
Spectrum index:
27: The number of concurrent illnesses such as epilepsy and intellectual disability affecting significantly more women than men with autism. The results appeared in the Journal of the American Medical Informatics Association in August.
336.6 percent: The increase in autistic adults receiving a US cash grant program in 2019 compared to 2005. The largest increases were in New Jersey, Delaware, and Connecticut, according to a study published in the Journal of Autism and Developmental Disorders August.
1.35: The increased likelihood that an autistic person’s cousin will have depression compared to someone who is not related to an autistic person. The result, published in the Journal of Child Psychology and Psychiatry in August, was based on data from nearly 2.4 million people in Sweden. Many other conditions, including anxiety, eating, and psychotic disorders, were also more common in relatives of people diagnosed with autism.
Dataset Spotlight: A new longitudinal dataset of the developing mouse brain, created using resting-state data on functional connectivity of the cortex, tracks the postnatal development of a number of animals over time. A preprint describing the data was published on bioRxiv in April.
Quote this article: https://doi.org/10.53053/QGXY6857
