AI and Alzheimer’s Disease: Pathway to Earlier and more Equitable Diagnoses

In a recent seminar presentation at Loyola University Chicago, Dr. Nicholas Baker discussed his research regarding the efficacy of deep convolutional neural networks (DCNNs) in matching human object recognition and visual processing patterns (Baker & Elder, 2022). The goal of the study was to assess DCNN image identification performance upon exposure to known and distorted shapes. Similarities in human and DCNN response accuracy would provide insight into whether DCNNs process visual information in the same way as the mammalian brain. The results of the study yielded several observations. First, humans and DCNNs both identified non-manipulated shapes with a high degree of accuracy, but demonstrated impaired identification performance for fragmented shapes. Humans and DCNNs also encountered “frankenstein” configurations, wherein the spatial arrangement of features were disrupted but the images remained whole. While humans had difficulty in this condition, DCNNs did not portray any deficiency. Accordingly, Dr. Baker explained that DCNNs rely on local shape cues, whereas humans integrate spatial information to derive overall meaning. This “small detail” versus “big picture” discrepancy highlights fundamental differences in how DCNNs and humans understand and interpret visual stimuli. 

Despite these findings, artificial intelligence technologies have proven useful in other areas of neuroscience; they may even have important healthcare implications. Two different studies, for instance, have investigated the use of AI as a diagnostic tool for dementia and Alzheimer's disease. In one randomized clinical trial, Boustani et al. paired the Quick Dementia Rating System (QDRS) with an AI passive digital marker, programmed to extract information from electronic health records (Boustani et al., 2025). In clinics that employed the QDRS and PDM, the rate of new Alzheimer’s diagnoses increased by 31%. Given that around 50% of adults do not receive official diagnoses, this study presents one way artificial intelligence can easily be used to improve early detection. 

A different team at UCLA has employed AI technologies to reduce racial bias in Alzheimer’s disease diagnoses. Prior research has revealed that diagnosis rates among non-white individuals fail to match up to higher relative prevalence rates, indicating that under-detection is pervasive in minority populations (Tran et al., 2025). The semi-supervised positive unlabeled learning (SSPUL) model was able to identify undiagnosed individuals with Alzheimer’s and demonstrated higher sensitivity than other AI technologies. Additionally, results provided by SSPUL were not altered when race and ethnicity data were changed, indicating low comparative racial bias. These findings are significant in regards to “catching” Alzheimer’s cases that would have otherwise gone undiscovered. 

Overall, what do these studies reveal about humans and AI? Dr. Baker’s research emphasizes that the human brain is enormously complex. We might still be quite far out from developing technology that accurately mirrors the complexity of human neural processing. Human propensities toward mistakes and bias, however, emphasize that there are benefits to not having AI think like us. As AI technologies continue to advance, they hold potential for massive healthcare implications, including Alzheimer’s diagnoses and beyond. AI technologies process information differently than humans, and thus have the capacity to fill in the gaps that humans have left open. 

References

Baker, N., & Elder, J. H. (2022). Deep learning models fail to capture the configural nature of human shape perception. iScience, 25(9), 104913. https://doi.org/10.1016/j.isci.2022.104913 

Boustani, M. A., Ben Miled, Z., Owora, A. H., Fowler, N. R., Dexter, P., Puster, E., Grout, R. W., Summanwar, D., Erazo, S. F., Disla, S., Coppedge, K., & Galvin, J. E. (2025). Digital detection of dementia in primary care. JAMA Network Open, 8(11). https://doi.org/10.1001/jamanetworkopen.2025.42222 

Tran, T., Fu, M., Fung, J., Sankararaman, S., Elashoff, D. A., Vossel, K., & Chang, T. S. (2025). Fair positive unlabeled learning for predicting undiagnosed Alzheimer's disease in diverse electronic health records. Npj Digital Medicine, 8(1). https://doi.org/10.1038/s41746-025-02111-1 

How to approach the ethical issues stemming from BCIs?

    In a past talk I had the opportunity to listen to a talk from Dr. Joe Vukov discussing the ethical implications of using brain computer interfaces on individuals that suffer from binge eating disorders. The article discussed was Brain-Responsive Neurostimulation for Loss of Control Eating: Early Feasibility Study (Wu et al.). Binge eating disorders are referred to as loss of control (LOC) eating, participants in the study have not succeeded with alternative treatment plans. The study experimented with delivering an electrical signal to interrupt the frequency that triggers LOC eating. In theory this can prevent the patient from being triggered to binge eat before the signal even reaches that part of their brain. This study opened a greater discussion on the ethics of brain computer interfaces, specifically the consent and personal control aspect.

    Yang and Jiang’s (2025) paper on Regulating neural data processing in the age of BCIs: Ethical concerns and legal approaches discusses the legal and ethical implications of brain computer interfaces (BCIs), and recommends legislative reform to protect patient privacy and rights.  As proposed in Wu’s study, if BCIs counteract signals when they are detected preventing them from reaching neural pathways, do patients still retain bodily autonomy? Wu’s research entails the implantation of a RNS system into the brain with the target region being the Nucleus Accumbens. When the anticipation of eating is detected an electrical signal is automatically delivered to the target region, the purpose being to interrupt the urge to binge eat. There is a window of vulnerability where the physiological changes are detectable, this is when the intervention is most opportune (Wu et al., line 25). Through this device the neural activity associated with urges, cravings, and emotional responses can be recorded. Data obtained through BCIs are a pathway into the neural patterns correlated with mental states raising ethical concerns about mental autonomy. Individuals do not often control what they are thinking, BCIs have the opportunity to record mental urges before an individual is consciously aware of them. This presents concerns in relation to free will and self control as the interventions can occur before conscious thought. Before 2024 there was no legislation in place to protect the data that is obtained from BCIs (Yang and Jiang, line 30-32). BCI data differs from biometric and genetic data because it is obtained through your mind, meaning it can record neural patterns associated with bias, intentions, thoughts and emotions. This type of collection is considerably more invasive than external choices made online. After 2024 states like Colorado and California have brief legislation vaguely protecting neural data, but this can be improved significantly. Neural data touches the brain in the most private part of the mind, thoughts, beliefs, and convictions are beyond what can be constrained by external forces (Yang and Jiang, line 20). BCIs having the ability to prevent these processes threatens individuals' mental autonomy. Having access to and record of these thoughts has the potential to be used unethically.

    Yang and Jiang’s article touches on the ethical implications of Wu’s research. Dr. Vukov presented similar concerns regarding patient protection when using BCIs. The use of BCIs for therapeutic treatment shows to be a promising technique, but the ethical implications must be considered. Legislation to protect patient rights should be advanced appropriately to prevent the misuse of data. The advancement of medical technology is important and should be utilized and in turn so should patient protections. 

References: 

Yang H, Jiang L. Regulating neural data processing in the age of BCIs: Ethical concerns and legal approaches. DIGITAL HEALTH. 2025;11. doi:10.1177/20552076251326123

Wu H, Adler S, Azagury DE, Bohon C, Safer DL, Barbosa DAN, Bhati MT, Williams NR, Dunn LB, Tass PA, Knutson BD, Yutsis M, Fraser A, Cunningham T, Richardson K, Skarpaas TL, Tcheng TK, Morrell MJ, Roberts LW, Malenka RC, Lock JD, Halpern CH. Brain-Responsive Neurostimulation for Loss of Control Eating: Early Feasibility Study. Neurosurgery. 2020 Nov 16;87(6):1277-1288. doi: 10.1093/neuros/nyaa300. PMID: 32717033; PMCID: PMC8599841.

Brain-Computer Interfaces and the Question of Autonomy

Brain-computer interfaces are systems that allow for communication between the brain and some other device, typically for the purpose of recording electric signals or restoring/altering functionality in a user. For example, they are often used for paralyzed individuals to use artificial limbs or to enable communication.

    Joe Vukov is a philosophy professor at Loyola University with an extensive knowledge of bio and neuroethics. He was recently able to give a talk surrounding brain-computer interfaces (BCIs) and their surrounding ethics. More specifically, he discussed a study in which neurostimulation was used to respond to loss of control eating in the article “Brain-Responsive Neurostimulation for Loss of Control Eating: Early Feasibility Study.” The study focuses on the use of a closed loop BCI in order to detect signals correlated with negative eating patterns prior to overeating and disrupt the process. The main concern regarding this form of treatment is whether or not patients will have true autonomy and consent over when these urges are controlled. Vukov pointed out that systems such as these could serve as an inhibitor of one’s decision making ability as patients are not fully able to stop treatment on a whim should they decide they would no longer wish to continue. Vukov went on to propose different ideas for future work with BCIs, such as those that could be controlled with a button for a patient to decide when they would like a response to be activated or not. 

In more advancements, researchers are working towards using these methods of stimulation for work related to strokes and neurodegenerative diseases. One article discussing such progress is “Advances in Brain–Computer Interfaces (BCI): Challenges and Opportunities”. The article goes into detail about proposed and ongoing work relating to these interfaces and what is in store for the future. For example, long-term use of BCI electrical stimulation has shown to increase neuroplasticity in older patients who experienced strokes. Another study mentioned in the article focused on tracking sleep staging.  BCIs are shifting in such a way that they are able to provide more modulatory impacts on neural circuitry over simply reading brain activity as well, allowing for more clinical uses of the technology. Researchers are also looking into further use for emotion recognition and disease detection. With all of these new advancements, the potential uses of these interfaces are endless.

The work being done by researchers is critical for the understanding of the possibilities for the clinical use of BCIs. From disease detection to enhancement of neuroplasticity, there are many directions in which this technology can be taken in order to aid those in need. While there is much work surrounding the abilities of these interfaces, there is less discussion on the ethical concerns that come with. Vukov’s discussion on privacy concerns and the autonomy of the patient is imperative in ensuring proper patient choice and understanding of these brain computer interfaces. While there are functions that may be critical, there still must be some level of autonomy for patients when deciding how these BCIs will function, and who all of the collected data will belong to or who has access to it. Vukov went in detail about potential motor discrimination and user-activation possibilities that may be implemented with design. With all the advances in clinical uses of BCIs, making sure that each user can maintain individuality and choice will be of utmost importance for researchers to consider in future work.

References:

Wu, H., Adler, S., Azagury, D. E., Bohon, C., Safer, D. L., Barbosa, D. A. N., Bhati, M. T., Williams, N. R., Dunn, L. B., Tass, P. A., Knutson, B. D., Yutsis, M., Fraser, A., Cunningham, T., Richardson, K., Skarpaas, T. L., Tcheng, T. K., Morrell, M. J., Roberts, L. W., Malenka, R. C., … Halpern, C. H. (2020). Brain-Responsive Neurostimulation for Loss of Control Eating: Early Feasibility Study. Neurosurgery, 87(6), 1277–1288. https://doi.org/10.1093/neuros/nyaa300

Wang, Y., Ge, M., & Xu, S. (2026). Advances in Brain-Computer Interfaces (BCI): Challenges and Opportunities. Biomimetics (Basel, Switzerland), 11(2), 157. https://doi.org/10.3390/biomimetics11020157 

Beyond the Symptoms: The Evolution of Using Sound to Detect Concussion in Children

Throughout the whole semester I had the overall pleasure to listen to Dr. Jennifer Kirzman speech on her research titled "Auditory biological marker of concussion in children", which centered around the impact concussion has on children's auditory skills, especially children from a sports medical clinic who had their concussions occur from a sports injury. This research highlights how concussions affect the auditory process in the Fundamental frequency (F0), and were ankle to explore if the concussion that children suffer affects the facilitation of pitch perception, identification of sound and talk, and the understanding of stress and prosody. This is what made me overall interested in the topic, since most of the time research around concussion in children focuses on brain-specific proteins as a biological marker. This research takes a different approach and focuses on the topic of auditory skills in order to diagnose children with concussions. 

One of the main concepts highlighted by the research is the shift in using auditory biological markers to diagnose concussion. The use of this biological marker is supported by the concept of examining how especially concussions impact neural processing in children. In the article "Persistent post-concussion symptoms include neural auditory processing in young children" the researchers highlight how children affected by consciousness display a disruption in their ability for their brains to encode sound, especially in processing pitch and speech-related frequencies. These types of disruption occur at the neural level, which signify that when a child's hearing ability may seem normal, their brains ability to accurately interpret the auditory information that is received is abnormal. This allows for the understanding of why children continue to struggle to follow conversations or understand speech in a loud environment even after they appear normal, or when a standard hearing test shows no worries.   Significantly, these types of changes can be consistent even after external symptoms amplify, proposing that auditory processing provides a more sensitive and unbiased measure of brain injury (concussions). 

Similarly, in the study titled "Long-term effects of mild traumatic brain injury in pediatrics", the research highlights how despite the fact that multiple children appear to recover within a short time a considerable number of children experience a continuous cognitive problem. This problem can include attention, learning, and memory. Significantly, these lasting effects are closely linked with systems that depend heavily on auditory processing. This connects to the aspect and setting that are important for children, classroom setting where students need the constant ability to interpret spoken languages and verbal instructions, all while filtering out background noises.  If concussions disrupt the children's ability on how the brain processes sound, it can also disrupt children's ability to learn and retain academically.  One example, a child's difficulty in encoding verbal lessons may make it hard for them to keep up with the teacher, even if they seem extremely fine. 

Together these research suggest that brain injury (such as concussions) cannot be completely understood only through visible symptoms or the traditional biological markers alone. Instead the use of examining long term effect and neural sound encoding, can show the use of auditory biomarkers as a shift to a more effective interpretation of concussion diagnosis and recovery. 

References
Bonacina, Silvia, et al. “Persistent Post-Concussion Symptoms Include Neural Auditory Processing in Young Children.” PubMed, vol. 9, no. 1, 1 Mar. 2024, pp. CNC114–CNC114, pmc.ncbi.nlm.nih.gov/articles/PMC11270634/, https://doi.org/10.2217/cnc-2023-0013.
Kraus, Nina, et al. “Auditory Biological Marker of Concussion in Children.” Scientific Reports, vol. 6, no. 1, Dec. 2016, www.nature.com/articles/srep39009, https://doi.org/10.1038/srep39009. Accessed 5 Dec. 2019.Tumarisi Tuersunjiang, et al. “Long-Term Effects of Mild Traumatic Brain Injury in Pediatrics.” Acta Psychologica, vol. 258, 15 July 2025, pp. 105260–105260, https://doi.org/10.1016/j.actpsy.2025.105260.

What Role Does Whole Shape(Configural Shape) Processing Play In Human Object Recognition Perception Systems and Why Do Artificial Neural Networks Do Not Succeed to Replicate It?

Nicholas Baker is a researcher in visual perception and computational neuroscience who has contributed a lot towards the understanding of how humans along with artificial intelligence recognize objects through shape. His work primarily focuses on the fundamental differences between human perception and deep learning models, specifically on how every system processes information on objects through shape. In the study, he expresses examination of how humans relied on the configural relationships between different parts of an object which means how the specific features are arranged relative to one another compared to only identifying isolated features local to the object(Baker et. al 2022). The concepts discussed in his research connect extremely closely to the ideas learned in class regarding perception and cognition where small changes in structure can have an impact significantly on recognition. Baker demonstrates that while humans naturally process objects as one whole, there are deep convolutional neural networks that often do not succeed to capture the configural nature, instead relying more heavily on local features or patterns. Their findings highlight that object recognition is not just primarily about identifying individual features of the object but about understanding the entire spatial relationships that bind those features into a whole. His research emphasizes that differences at the smallest levels for visual organization can have extremely significant impact on how both biological and artificial systems interpret the objects in the world.  

According to the study conducted by Nicholas Baker, human object perception relies heavily on the configural relationships between the local shape components of the object, which means that individuals recognize objects not just by isolated parts but by how those parts are spatially arranged into a coherent whole(Baker et. al 2022). In their experiments, participants were shown animal silhouettes that were either intact, broken, or rearranged referenced as the Frankenstein condition while also preserving local features on the objects. The results displayed and expressed that disrupting the overall configuration significantly causes problems with human recognition performance in which they decline even when the specific features on the object stay the same. In contrast, the deep convolutional neural networks (DCNNs), such as ResNet-50 and VGG-19 displayed little to no decrease in the performance when configural relationships were disrupted which helped indicate that the following models rely much on the local shape cues(or features) comparatively to processing as a whole(holistically). Furthermore, even when networks were retrained to emphasize shape over texture, or adjusted with more complex architectures such as recurrent connections in which they still did not succeed to exhibit human-like configural sensitivity(Baker et. al 2022). As a result, the study concludes that while humans process visual information in a whole perspective and spatially integrated manner, current deep learning models that lack the ability and instead function similarly to the systems that add up independent components in which it highlights a fundamental difference between biological and artificial vision systems that suggest future models must primarily incorporate border perceptual tasks for better replication experiments on the configural nature of human shape perception. 

Human perception of visual objects relies strongly on the ability to organize individual components into a whole which is a process that is essential for accurate recognition. A recent study by Dehn et al. investigated to recognize objects even when visual information is incomplete or altered. The researchers expressed that humans do not simply detect isolated parts of an object but primarily rely on the relationships between those parts toward a complete perception of shape. As a result, it suggests that the brain naturally prioritizes whole structure and configural relationships  compared to individual features of the object during visual processing of the objects(Dehn et. al 2025). However, in the study there is no direct examination of deep learning models, the findings are extremely consistent with those from Baker’s study in which together they highlight the fundamental differences in visual processing with humans compared to artificial intelligence. Baker displayed that humans cannot process configural relationships when they are altered. Deep neural networks remain largely unaffected which indicates a reliance on local features compared to whole processing. The similarity between these studies suggests that human perception is inherently structured around the spatial relationships between the individual features, whereas current artificial systems do not succeed to capture that level of organization(Dehn et. al 2025). However, it raises an important question: if human perception depends primarily on whole image processing, could future machine learning models be designed to process images via configural processing leading to more effective human-like visual recognition based technological systems? 

The ability to organize and interpret visual information through using separate parts then combining them to form the whole image has a significant impact on how humans understand the world such as when driving and you notice a stop sign based on the red octagonal shape without seeing the word stop.  The research from Baker demonstrates that human object recognition depends significantly on configural processing while the study from Dehn further supports that the brain naturally puts the small pieces together to form the entire image. Together both studies display that human perception is based on seeing the entire picture and not the different parts that make up the whole image. However, most artificial intelligence systems do not process the images based on a whole of parts, they focus primarily on the small, minute details. The difference expresses that understanding how humans process visual stimuli and information can help improve artificial intelligence programs in the future in accurately processing visual stimuli similarly to humans. 

References 

Baker , Nicholas, and James  H Elder. Deep Learning Models Fail to Capture the Configural Nature of Human Shape Perception, 16 Sept. 2022, www.sciencedirect.com/science/article/pii/S2589004222011853. 

Dehn, Kira  Isabel, et al. “Human Shape Perception Spontaneously Discovers the Biological Origin of Novel, but Natural, Stimuli | Journal of the Royal Society Interface | The Royal Society.” Human Shape Perception Spontaneously Discovers the Biological Origin of Novel, but Natural, Stimuli, 21 May 2025, royalsocietypublishing.org/rsif/article/22/226/20240931/235863/Human-shape-perception-spontaneously-discovers-the.

The LDT to VTA Pathway and Cocaine Sensitization

The laterodorsal tegmental nucleus (LDT) is a brainstem region located in the pontine tegmentum - located in the very rear part of the pons. The LDT contains glutamatergic neurons and plays a crucial role in modulating mood and attention as well as regulating sleep and the development of REM sleep. The ventral tegmental area is also a part of the brainstem but is found in the midbrain. This region contains mostly dopaminergic neurons projecting to the nucleus accumbens and prefrontal cortex and works to control reward and addiction (Cai et al, 2022). More recent studies have shown that the LDT actually does more than regulate mood, attention, and sleep. It also works with the VTA to mediate reward and addiction. The LDT sends signals, or glutamatergic projections, that excite the VTA dopaminergic neurons leading to reward-related behaviors (Coimbra et al., 2021) and even cocaine sensitization. 

In the research paper, “Glutamate inputs from the laterodorsal tegmental nucleus to the ventral tegmental area are essential for the induction of cocaine sensitization in male mice” by Dr. Amit Puranik et al., they further analyze the LDT-VTA pathway in the brain to investigate the specific role of the glutamatergic neurons as well as the surrounding sensory neurons in cocaine sensitization. In the experiment, they utilized optogenetics in male mice to selectively inhibit the glutamatergic neuronal projections as they repeatedly injected cocaine to promote sensitization. Typically, the mice would show increased movement, but the researchers observed that the inhibition of glutamate did not result in the typical increase of movement in these mice proving that glutamate signaling from the LDT to the VTA is crucial for cocaine sensitization because it drives the development of the VTA synapses known to enhance the addictive behaviors (Puranik et al., 2022). Based on their results, the researchers concluded that constant exposure to cocaine strengthens the synapses between glutamatergic and dopaminergic neurons leading to the sensitization that was not observed when the glutamatergic neuron projections were inhibited. This not being directly tested and only inferred is one of the few limitations of this article. The researchers did not directly measure plasticity but instead measured behavioral outcomes and dopamine release. 

In the article, “Drug-Evoked Synaptic Plasticity of Excitatory Transmission in the Ventral Tegmental Area” by Camilla Bellone and colleagues, they directly address synaptic plasticity as a result of addictive drugs. Although this article is a review article, it synthesizes key techniques used that directly measure synaptic strength rather than solely relying on behavioral and chemical outcomes. The article mentions that there is a significant increase in the AMPA  and NMDA ratio as a result of drug exposure (Bellone et al, 2021). Both receptors are glutamate binding and key indicators of synaptic plasticity. During long term potentiation from an addictive drug, there is a significant increase in the number of AMPA receptors on the postsynaptic dopaminergic neuron. This results in stronger synapses between the glutamatergic and dopaminergic neuronal projections leading to sensitization, when dopamine neurons then become more responsive and lead to greater behavioral response to the addictive drug.    

Both articles work together to prove that behavioral sensitization is very complex. The articles link observable behavior to the cellular mechanisms within the nervous system that causes these behaviors. They both come together to explain how sensitization causes neuroplasticity. 

Works Cited

Cai, J., & Tong G. (2022). Anatomy and Function of Ventral Tegmental Area Glutamate Neurons. Frontiers in neural circuits, 16, 867053. https://doi.org/10.3389/fncir.2022.867053

Coimbra, B., Domingues, A. V., Soares-Cunha, C., Correia, R., Pinto, L., Sousa, N., & Rodrigues, A.J. (2021). Laterodorsal tegmentum-ventral tegmental area projections encode positive reinforcement signal. Journal of neuroscience research, 99(11), 3084-3100. https://doi.org10.1002/jnr.24931

Bellone, C., Loureiro, M., & Lüscher, C. (2021). Drug-Evoked Synaptic Plasticity of Excitatory Transmission in the Ventral Tegmental Area. Cold Spring Harbor perspectives in medicine, 11(4), a039701. https://doi.org/10.1101/cshperspect.a039701

Puranik, A., Buie, N., Arizanovska, D, Vezina, P., & Steidl, S. (2022). Glutamate inputs from the laterodorsal tegmental nucleus to the ventral tegmental area are essential for the induction of cocaine sensitization in the male mice. Pyschopharmacology, 239(10), 3263-3276. https://doi.org//10.007/s00213-022-06209-2

Teenage Sleep vs. Its Biggest Impact – Cellphones

    It’s 1 am, and you were supposed to be asleep for your exam at 8 am the next morning. The artificial sun is pouring into your eyes in your dark room. After noticing the time, you decide to go to sleep, but it takes you an hour. The next morning, you walk into your class groggy, and you end up passing the exam with a C. Not the grade you wanted or studied for.

    I recently attended a seminar by Dr. Stephanie Crowley, who talked about her article about adolescent sleep called “An update on adolescent sleep: New evidence informing the perfect storm model”. In this seminar, she mentioned that there are two internal systems that control sleep. In adolescence, sleep is naturally pushed later, but things such as societal pressure require early wake-up times. The homeostatic system that builds pressure to sleep builds more slowly than in childhood. This results in the internal clock, circadian rhythm, shifting later, causing drowsiness to start later. Falling asleep later causes waking up later to get the required nine hours of sleep. This is a problem since school requires students to be at school before 8 am, and if students don’t get tired earlier, they won’t be able to get enough sleep. Along with this, there is light from screens. This light causes your brain not to produce melatonin.

    In a 2023 literature review called “The Influence of Smartphones on Adolescent Sleep: A Systematic Literature Review,” the researchers reviewed how smartphones specifically influenced sleep quality in adolescents. They found that adolescents who have a smartphone sleep fewer hours compared to those without one. 97% of teenagers involved in a study used some screen before bed, with the most common device being a smartphone. The activities on these devices such as texting or scrolling on social media keeps the brain active. Along with that, the blue light emitted from a screen reduces melatonin secretion, causing a later sleep time. Multiple researchers have found a similar conclusion. Smartphones cause cognitive arousal during a time that the brain should be winding down. A bad night’s sleep has been linked to poor sleep, depressive mood, diminished coping abilities, and reduced academic performance.

    This links directly to Dr. Crowley’s talk about the “perfect storm.” Not getting the right amount of sleep has a ripple effect on the rest of life. There are multiple factors that combine to make adolescent sleep worse. Smartphone usage before bed keeps the brain up and doesn’t let it do what it has done for thousands of years. School requires teenagers to wake up early, while they fall asleep later due to biology. As a society, there would need to be shifts to better help and fit the adolescent sleep schedule. Adolescents can also help themselves by not using phones before bed and starting the natural cycle of sleep.

References

Crowley, Stephanie J., et al. “An Update on Adolescent Sleep: New Evidence Informing the Perfect Storm Model.” Journal of Adolescence, vol. 67, no. 67, Aug. 2018, pp. 55–65, https://doi.org/10.1016/j.adolescence.2018.06.001.

Sofia de Sá, et al. “The Influence of Smartphones on Adolescent Sleep: A Systematic Literature Review.” Nursing Reports, vol. 13, no. 2, Apr. 2023, pp. 612–21, https://doi.org/10.3390/nursrep13020054.

Brain-Computer Interfaces: Privacy, Consent, and Control

    In his recent talk, Joe Vukov explored brain-computer interfaces and their ethical implications, focusing mainly on issues of autonomy and privacy. He discussed a study by Hemmings Wu et al. on a closed-loop brain-computer interface (BCI) designed to treat loss of control eating in patients for whom previous treatment had been unsuccessful. The system detects neural signals associated with unhealthy eating behaviors originating in the nucleus accumbens and responds by delivering deep brain stimulation via the RNS System, preventing these impulses from developing into physical actions. As a professor of philosophy, Vukov’s main concern was with the autonomy and privacy retained by patients who opted in to treatments such as these. Because the system operates in a closed loop, its effects are not externally visible, making it difficult for patients to monitor, evaluate, or end the treatment once it begins. As Vukov discussed, this type of automated control risks limiting a patient’s ability to consent in real time or withdraw from treatment, drawing attention to the decrease in direct control that users will have as BCIs become more effective. He contrasts this with open-loop systems that translate the neural signals into external actions, such as speech or movement of a limb, introducing the concern that private thoughts could be continuously monitored or exposed without sufficient user control.

    In a new study, Erin M. Kunz et al. investigated the use of BCIs to decode inner speech in individuals with paralysis as a means to restore communication. For this study, four participants from the BrainGate2 trial were recruited, each with varying abilities to produce speech or communicate with others. In order to study the neural representations of their speech, microelectrode arrays were placed in the precentral gyrus of each participant. This area of the motor cortex produces neural activity associated with inner speech, perceived speech, and reading. After performing various tasks, some of the participants’ results showed a decoding accuracy for inner and perceived speech that was the same or better than attempted speech. It was also found that the neural representations of attempted and inner speech overlapped, showing a correlation in neural firing rates. This showed that words are encoded similarly across different behaviors, but also raises the possibility of unintentionally decoding the private inner speech of users. To address this concern, the researchers shifted focus to the difference in motor-intent signal between inner and attempted speech, which allowed decoders to distinguish between the behaviors and more accurately transcribe the desired speech. Another proposed solution to prevent unwanted decoding of inner speech is to employ a user-controlled keyword that could be said by users to “lock” and “unlock” the decoders, allowing inner speech to continue without being expressed out loud.

    The work done by Wu et al. and Kunz et al. highlight a promising direction for BCIs to continue making meaningful impacts on patients’ lives, while still preserving their privacy. Vukov’s discussion of BCIs show that systems that work independently of user acknowledgement risk diminishing user autonomy, regardless of therapeutic benefits. The research done by Kunz et al., however, demonstrates that it is possible for BCIs to be designed with user intention and control as a priority. By incorporating preventative measures such as motor-intent discernment or user-activated keywords, boundaries between private thoughts and deliberate speech can be established. These measures address the concerns raised by Vukov, suggesting that the ethical risks of BCIs are not necessarily a symptom of the technology, but rather, dependent on the design and implementation of each treatment system. Ensuring that users retain control over when and how their neural activity is interpreted will be essential for the further development and use of BCIs.

References:

Kunz, E. M., Abramovich Krasa, B., Kamdar, F., Avansino, D. T., Hahn, N., Yoon, S., Singh, A., Nason-Tomaszewski, S. R., Card, N. S., Jude, J. J., Jacques, B. G., Bechefsky, P. H., Iacobacci, C., Hochberg, L. R., Rubin, D. B., Williams, Z. M., Brandman, D. M., Stavisky, S. D., AuYong, N., … Willett, F. R. (2025). Inner speech in motor cortex and implications for speech neuroprostheses. Cell, 188(17). https://doi.org/10.1016/j.cell.2025.06.015 

Wu, H., Adler, S., Azagury, D. E., Bohon, C., Safer, D. L., Barbosa, D. A. N., Bhati, M. T., Williams, N. R., Dunn, L. B., Tass, P. A., Knutson, B. D., Yutsis, M., Fraser, A., Cunningham, T., Richardson, K., Skarpaas, T. L., Tcheng, T. K., Morrell, M. J., Roberts, L. W., Malenka, R. C., … Halpern, C. H. (2020). Brain-Responsive Neurostimulation for Loss of Control Eating: Early Feasibility Study. Neurosurgery, 87(6), 1277–1288. https://doi.org/10.1093/neuros/nyaa300

Can Artificial Intelligence Replace Humans in Histopathology?

    Convolutional neural networks (CNNs) have been an important component in the development of artificial intelligence (AI) and machine learning. It has been used in interpreting visual and spatial data and is being trained to perform in tasks related to image recognition, language processing, and tasks in the medical field (ScienceNewsToday). Delving into the medical field, pathology is a specialty that involves histology, which means studying tissue samples under a microscope to determine if the tissue is diseased. As artificial intelligence such as CNNs continue to evolve, it begs the question, can they replace humans in histopathology?

    During Dr. Baker’s talk at Loyola, he highlighted his own research regarding how deep convolutional models (DCNNs) struggle with the configuration of images compared to humans. In his paper, Dr. Baker highlighted how DCNNs focused more on color and texture than humans do (Baker). Additionally, in his presentation he went into detail into how the AI struggled with recognizing an image based on manipulation of the images’ border. For example, if the AI was given the silhouette of a cat, but if the border was turned into rigid edges rather than the normal “smooth” border, it would have more trouble recognizing what the image was in comparison to an image that was “Frankensteined” (chopped up and put together in a different configuration).

    It is necessary to make note of this distinction in the AI’s recognition of the image because cell tissues do not always maintain a constant shape or look, meaning that Dr. Baker findings could pose an issue of accuracy when using AI in pathology. A review article by Prasad et al. delves further into the use of AI in the field of histopathology. In their findings, CNNs were found to have consistent results in detecting and classifying cancerous tissue and had levels of accuracy similar to trained pathologists in controlled settings (Prasad). However, the article highlights an issue in which AI systems struggle with variability in staining, slide preparation, and tissue morphology (Prasad).

    This connects to Dr. Baker’s findings, as the AI relies heavily on texture and color rather than truly understanding structural features, which limits the effectiveness of CNNs when analyzing histological samples that are inconsistent or irregular (Baker). Furthermore, Prasad et al. highlight that while AI can assist with lowering workloads and increasing efficiency, as mentioned, it still lacks the adaptability needed to analyze irregularities that can be common in complex histopathological cases (Prasad). This shows that while AI can become an assistive tool in histopathology, its limitations in image interpretation demonstrate that humans in pathology are still necessary. However, rather than replacing pathologists, AI is more useful as a supportive tool that enhances accuracy and efficiency.

 

References

Editors of ScienceNewsToday. (2026, April 7). Convolutional neural networks: the science behind modern artificial intelligence. Science News Today. https://www.sciencenewstoday.org/convolutional-neural-networks-the-science-behind-modern-artificial-intelligence#google_vignette

Baker N, Elder JH. Deep learning models fail to capture the configural nature of human shape perception. iScience. 2022 Aug 11;25(9):104913. doi: 10.1016/j.isci.2022.104913. PMID: 36060067; PMCID: PMC9429800.

Prasad P, Khair AMB, Saeed M, Shetty N. Artificial Intelligence in Histopathology. J Pharm Bioallied Sci. 2024 Dec;16(Suppl 5):S4226-S4229. doi: 10.4103/jpbs.jpbs_727_24. Epub 2025 Jan 30. PMID: 40061791; PMCID: PMC11888715.

Peripheral Nerve Degeneration: How the Shock Affects Nerve Repair

     In science, there has been an uptick in the use of electrical stimulation to "biohack" our bodies and brains, whether it is to wake us up or for physical therapies. As a result, many new studies have emerged to understand the body's natural stimulation and create tech that mimics it to address a variety of issues. Currently, scientists are digging deeper to determine which frequencies are the best for humans to create devices specifically for nerve degeneration.   

         During our class, we heard from Dr. Vincent Chen, whose focus was Power Spectral Density (PSD). PSD describes the power of a time-domain signal, or how a random process can be distributed across different frequencies. Chen's hypothesis is as follows: we should manipulate the waveform shape to target NMDA and AMPA receptors, which act as switches that turn on nerve growth. His clinical success with "random noise" stimulation is due to the rich PSD. He also questions modern electrical stimulation methods, as square waves commonly used contain harmonics, echoes of higher frequencies hidden within the signal. With these higher frequencies, it makes it harder to determine which signal is most optimal for nerve regeneration actively. Just following the frequency to him is imprecise because the signal will degrade as soon as it passes through the skin and tissue. He notes that the nerve membrane acts as a capacitor, resisting sudden changes, so changing the type of wave could also affect the nerve differently.  By controlling the voltage gradient, he can target very specific receptors. Chen's approach is highly specific, and with the future of implantable devices, it blows all past nerve studies out of the water.   

A different study conducted by Dr. Lingmei Ni uses very traditional electrical stimulation approaches to alleviate the effects of nerve damage. Ni uses multiple types of electrical stimulation, including NMES (neuromuscular), and low-frequency pulses to contract the muscles, helping prevent muscle atrophy directly. TENS is transcutaneous, blocking the pain signals with varying high and low frequencies. FES is the functional frequency, which can help paralyzed limbs return to function. These therapies have been used for a long time, but Ni has clinical data on specific frequencies to support her statement. Her studies show that 1 hour of 20Hz stimulation is proficient to accelerate axon growth after carpal tunnel surgery. Electrical stimulation increases BDNF and cAMP, which act as fuel for a growing neuron. However, even Dr. Ni points out how there is no standard for electrical stimulation. It varies between patients, parts of the body, and at times can seem almost random.   

These studies, when read together, can be seen as the future of electrical stimulation. Dr. Ni has a wide breadth of knowledge and clinical data on electrical stimulation, helping regenerate axons faster. Dr. Chen takes it one step further; no longer will the frequency or even type of wave vary from person to person. Rather, using PSD data, we can build stimulators that do not cause nerve fatigue or unwanted pain from high-frequency harmonies. With this amalgamation of information, other scientists can take this data to build more biohacking devices to help humans live more comfortable lives following nerve degeneration.   

Chen, VincentC.-F., et al. "Accelerating Peripheral Nerve Regeneration Using Electrical  Stimulation of Selected Power Spectral Densities." Neural Regeneration Research,  vol. 17, no. 4, 2022, p. 781, https://doi.org/10.4103/1673-5374.322458.  

Ni, Lingmei, et al. "Electrical Stimulation Therapy for Peripheral Nerve Injury." National  Center for Biotechnology Information, U.S. National Library of Medicine, 23 Feb.  2023, pmc.ncbi.nlm.nih.gov/.