Neuromorphic computing for modeling neurological and psychiatric disorders: implications for drug development

Artificial Intelligence Review, Год журнала: 2024, Номер 57(12)

Опубликована: Окт. 10, 2024

Abstract The emergence of neuromorphic computing, inspired by the structure and function human brain, presents a transformative framework for modelling neurological disorders in drug development. This article investigates implications applying computing to simulate comprehend complex neural systems affected conditions like Alzheimer’s, Parkinson’s, epilepsy, drawing from extensive literature. It explores intersection with neurology pharmaceutical development, emphasizing significance understanding processes integrating deep learning techniques. Technical considerations, such as circuits into CMOS technology employing memristive devices synaptic emulation, are discussed. review evaluates how optimizes discovery improves clinical trials precisely simulating biological systems. also examines role models comprehending disorders, facilitating targeted treatment Recent progress is highlighted, indicating potential therapeutic interventions. As advances, synergy between neuroscience holds promise revolutionizing study brain’s complexities addressing challenges.

Язык: Английский

---

Best practices for clinical trials of deep brain stimulation for neuropsychiatric indications

Frontiers in Human Neuroscience, Год журнала: 2025, Номер 19

Опубликована: Апрель 16, 2025

Deep brain stimulation (DBS) is well suited to target disorders with network dysregulation, as the case in many neuropsychiatric diseases. While DBS a well-established therapy for Parkinson's disease, essential tremor, dystonia, and medically refractory epilepsy, it actively being studied clinical trials including treatment-refractory major depressive disorder (MDD). Due nature of symptomology participant characteristics, special care must be taken design implementation testing disorders. In particular, these studies typically include multi-year relationships between participants study staff frequent interactions, high burden activities on participants, disclosure by sensitive information related symptoms disease state. Through our experience six across more than 5 years Presidio trial assessing personalized closed-loop MDD, we have gathered evidence inform best practices conducting interaction-intensive vulnerable population. Here, present Key Practices along discussion, informed multiple fundamental principles: The Belmont Report; emotional physical safety staff; integrity validity scientific outcomes.

Язык: Английский

---

Will adaptive deep brain stimulation for Parkinson’s disease become a real option soon? A Delphi consensus study

npj Parkinson s Disease, Год журнала: 2025, Номер 11(1)

Опубликована: Май 5, 2025

While conventional deep brain stimulation (cDBS) treatment delivers continuous electrical stimuli, new adaptive DBS (aDBS) technology provides dynamic symptom-related stimulation. Research data are promising, and devices already available, but we ready for it? We asked leading experts worldwide (n = 21) to discuss a research agenda aDBS in the near future allow full adoption. A 5-point Likert scale questionnaire, along with Delphi method, was employed. In next 10 years, will be clinical routine, is needed define which patients would benefit more from treatment; second, implantation programming procedures should simplified actual generalized adoption; third, algorithms, integration of paradigm technologies, improve control complex symptoms. Since years crucial implementation, focus on improving precision making accessible.

Язык: Английский

---

The Epistemological Consequences of Artificial Intelligence, Precision Medicine, and Implantable Brain-Computer Interfaces

Voices in Bioethics, Год журнала: 2024, Номер 10

Опубликована: Июнь 30, 2024

ABSTRACT I argue that this examination and appreciation for the shift to abductive reasoning should be extended intersection of neuroscience novel brain-computer interfaces too. This paper highlights implications applying personalized implantable neurotechnologies. Then, it explores whether is sufficient justify insurance coverage devices absent widespread clinical trials, which are better applied one-size-fits-all treatments. INTRODUCTION In contrast classic model randomized-control often with a large number subjects enrolled, precision medicine attempts optimize therapeutic outcomes by focusing on individual.[i] A recent publication strengths weakness both traditional evidence-based medicine.[ii] Plus, outlines tension in from medicine’s inductive style (the collection data postulate general theories) generation an idea limited available).[iii] The paper’s main example application treatment cancer.[iv] As name suggests, significant advancement neurotechnology directly connects someone’s brain external or implanted devices.[v] Among various kinds interfaces, adaptive deep stimulation require numerous adjustments their settings during implantation computation stages order provide adequate relief patients treatment-resistant disorders. What makes these unique how integrates sensory component initiate stimulation. While not commonly at level sophistication as self-supervising generative language models,[vi] they currently allow semi-autonomous form neuromodulation. treatments.[vii] ANALYSIS I. State Precision Medicine Oncology Epistemological Shift thorough overview cancer beyond scope article, its practice can roughly summarized identifying clinically characteristics patient possesses (e.g., genetic traits) land specialized option that, theoretically, benefit most.[viii] However, such stratification fall into smaller populations quality evidence anyone outside decreases turn.[ix] logic helps articulate, greater respond particular therapy higher probability efficacy. By straying logical framework, opens more uncertainty about validity approaches resulting disease subcategories.[x] Thus, while contemporary medical practices explicitly describe some treatments “personalized”, ought viewed inherently founded than other therapies.[xi] relevant case out Norway focuses care between ventricles heart esophagus, had failed standard regimen therapies over four years.[xii] last-ditch effort, elected pay out-of-pocket experimental immunotherapy (nivolumab) private hospital. He experienced marked improvements reduction size tumor. Understandably, tried pursue further rounds nivolumab public hospital initially declined given “lack randomised trials drug relating [patient’s] condition.”[xiii] rebuttal claim, countered he was actually similar subpopulation who responded “open‐label, single arm, phase 2 studies another immune drug” (pembrolizumab).[xiv] Given interpretation prior patient’s response, were approved. Had tumor’s following round nivolumab, then pembrolizumab’s empirical isolation would have been insufficient, inductively speaking, his continued use nivolumab.[xv] demonstrates induction abduction. phenomenon ‘cancer improvement’ considered causally linked underlying physiological mechanisms.[xvi] “the abductions there may always better, unknown explanation effect. belong special subgroup spontaneously improves, change placebo does mean, however, inferences cannot strong reasonable, sense make conclusion probable.”[xvii] To demonstrate limitations relying isolation, commentators pointed side effects hard rule being related initial intervention itself unless trends group taken consideration.[xviii] artificial intelligence (AI) assists development oncology, consideration. implementation AI has crucial providing way combine datasets variables machine learning recommend matches based statistics success upon practitioners base recommendations.[xix] usually establishing causal relationship[xx] – predicting. So, bleeds devices, like same cautions using alone carried over. II. Responsive Neurostimulation, AI, Personalized Like treatment, computer-brain interface technology similarly individual through settings. properly expose medicine, reasoning, neurotechnologies, descriptions systems need deepen.[xxi] broad summary stimulation, neural signal, typically referred local field potential,[xxii] must first detected interpreted device. device premarket approval, NeuroPace Neurostimulation system, used treat epilepsy detecting storing “programmer-defined phenomena.”[xxiii] Providers detection align electrographic seizures well personalize reacting stimulation’s parameters.[xxiv] provider adjusts trial error. One day algorithms will able regularly aid process myriad ways, specific ahead time electrophysiological signatures.[xxv] Either way, programmers, neurostimulation technologies individualized therefore operate line rather trials. Contemporary sophisticated enough prominent discussions where topics networks, learning, models, self-attention dominate conversation. high-density electrocorticography arrays (a much sensitive version use) combination networks help neurologic deficits stroke “speak” virtual avatar.[xxvi] situations, optimizing parameters increasing levels independence.[xxvii] An analogous surrounds United States experiencing OCD temporal lobe epilepsy.[xxviii]Given refractory nature her epilepsy, system indicated. therapy, also indicated off-label set-up. Another lead, one placed right nucleus accumbens ventral pallidum region correlation nuclei symptoms research. Following this, underwent “1) ambulatory, patient-initiated magnet-swipe storage moments obsessive thoughts; (2) lab-based, naturalistic provocation OCD-related distress (naturalistic task); (3) VR [virtual reality] (VR task).”[xxix] Such signals identify when deliver counter symptoms. Thankfully, procedure calibration exhibited recently shared results publicly.[xxx] cases, justification efficacy delivered therapy. study treated least activity tested determine optimum avoid them guesswork. Additionally, lead already before conducted, meaning bulk procedural risk could determined. test replicated biopsied against remaining immunotherapies vitro. Yet, few options, previous dose appeared work doses. Norwegian presents, corroboration known responses (from trial) helpful validate strategy. (It noted resigned last resort options regardless treatment.) There elements seen research general. For example, abductively focus X’s different Y’s Z’s. contrast, grouped obtained X, Y, Z aspect approach’s safety and/or holds plenty approach treating individuals try method, additional data. With gradual integration efficacy, reliance abduction continue, if grow, time. Moving forward, responsive (like nivolumab) suggestion similarities literature), investigative intervention, unrelated reasons deny it. III. Ethical Implications Next Steps AI’s oncology neurology yet fields radiology), appears horizon both.[xxxi] found functioning neurotechnologies medicine. serve individualize oncologic neurological therapies. handful publications cited important nuanced evaluation treatments, heavily rely justification, managed. just difficult infused pursued. At baseline, relies advanced literacy among exclude lack access basic technological infrastructure know-how participation.[xxxii] Even nations infrastructure, seek robust healthcare resources, market favor afford complex care.[xxxiii] If means dose/use product pocket, providers required cover subsequent treatments?[xxxiv] That is, stimulator battery life successful, feel justified having costs covered. experience implies precedent companies successful therapies, all see themselves obligated precision/abductive CONCLUSION fact cases outlined above insurance, individualized, compared groups standardized protocol (settings/doses). examining cohort groups/phases, conclude symptom likely coming themselves. preference take priority ruling funding neurostimulator. nuances discussion surrounding classifications interventions versus warrant future exploration, since distinction scale[xxxv] binary impacts “right-to-try” States.[xxxvi] Namely, inherent conducting neuropsychiatric disorders, surgically innovative frameworks blend methodologies, sham phases, traditionally used.[xxxvii] Similarly, systems, no instead only something worked someone else, then, addition treatment/dose question, balance valid arguably coverage. become common, evaluating decision making. ACKNOWLEDGEMENT article originally written assignment Dr. Francis Shen’s “Bioethics & AI” course Harvard’s Center Bioethics. thank Shen comments my colleagues Lázaro-Muñoz Lab fo - [i] Jonathan Kimmelman Ian Tannock, “The Paradox Medicine,” Nature Reviews. Clinical 15, no. 6 (June 2018): 341–42, https://doi.org/10.1038/s41571-018-0016-0. [ii] Henrik Vogt Bjørn Hofmann, “How Changes Medical Epistemology: Formative Case Norway,” Journal Evaluation Practice 28, (December 2022): 1205–12, https://doi.org/10.1111/jep.13649. [iii] David Barrett Ahtisham Younas, “Induction, Deduction Abduction,” Evidence-Based Nursing 27, 1 (January 1, 2024): 6–7, https://doi.org/10.1136/ebnurs-2023-103873. [iv] Epistemology,” 1208. [v] Wireko Andrew Awuah et al., “Bridging Minds Machines: Recent Advances Brain-Computer Interfaces Neurological Neurosurgical Applications,” World Neurosurgery, May 22, 2024, S1878-8750(24)00867-2, https://doi.org/10.1016/j.wneu.2024.05.104. [vi] Mark Riedl, “A Very Gentle Introduction Large Language Models without Hype,” Medium (blog), 25, 2023, https://mark-riedl.medium.com/a-very-gentle-introduction-to-large-language-models-without-the-hype-5f67941fa59e. [vii] E. Burdette Barbara Swartz, “Chapter 4 Neurostimulation,” Epilepsy, ed. Vikram R. Rao (Academic Press, 2023), 97–132, https://doi.org/10.1016/B978-0-323-91702-5.00002-5. [viii] 2018. [ix] [x] Simon Lohse, “Mapping Uncertainty Medicine: Systematic Scoping Review,” 29, 3 (April 2023): 554–64, https://doi.org/10.1111/jep.13789. [xi] Medicine.” [xii] 1206. [xiii] [xiv] [xv] 1207. [xvi] [xvii] [xviii] 1210. [xix] Mehar Sahu Three Artificial Intelligence Machine Learning Paradigm Big Data Analysis,” Progress Molecular Biology Translational Science, B. Teplow, vol. 190, vols., 2022), 57–100, https://doi.org/10.1016/bs.pmbts.2022.03.002. [xx] Stefan Feuerriegel “Causal Predicting Treatment Outcomes,” 30, 958–68, https://doi.org/10.1038/s41591-024-02902-1. [xxi] Sunderland Baker “Ethical Considerations Closed Loop Deep Brain Stimulation,” Stimulation (October 8–15, https://doi.org/10.1016/j.jdbs.2023.11.001. [xxii] Haslacher “AI Interfaces,” 7, https://doi.org/10.1016/bs.dnb.2024.02.003. [xxiii] 103–4; “Premarket Approval (PMA),” https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpma/pma.cfm?id=P100026. [xxiv] 104. [xxv] 126. [xxvi] Sean L. Metzger High-Performance Neuroprosthesis Speech Decoding Avatar Control,” 620, 7976 (August 1037–46, https://doi.org/10.1038/s41586-023-06443-4. [xxvii] Hao Fang Yuxiao Yang, “Predictive Neuromodulation Cingulo-Frontal Neural Dynamics Major Depressive Disorder Using Interface System: Simulation Study,” Frontiers Computational Neuroscience 17 (March 6, https://doi.org/10.3389/fncom.2023.1119685; Mahsa Malekmohammadi “Kinematic Adaptive Resting Tremor Parkinson’s Disease,” Movement Disorders 31, (2016): 426–28, https://doi.org/10.1002/mds.26482. [xxviii] Young-Hoon Nho “Responsive Guided Ventral Striatal Electrophysiology Obsession Durably Ameliorates Compulsion,” Neuron 0, 0 20, https://doi.org/10.1016/j.neuron.2023.09.034. [xxix] al. [xxx] al.; Erik Robinson, “Brain Implant OHSU Successfully Controls Both Seizures OCD,” News, accessed March 3, https://news.ohsu.edu/2023/10/25/brain-implant-at-ohsu-successfully-controls-both-seizures-and-ocd. [xxxi] Machines”; Interfaces.” [xxxii] Machines.” [xxxiii] Sara Green, Prainsack, Maya Sabatello, Roots (in)Equity Gaps Discourse,” 21, 5–9, https://doi.org/10.2217/pme-2023-0097. [xxxiv] 7. [xxxv] Robyn Bluhm Kirstin Borgerson, “An Epistemic Argument Research-Practice Integration Philosophy: Forum Bioethics Philosophy 43, (July 9, 469–84, https://doi.org/10.1093/jmp/jhy009. [xxxvi] Vijay Mahant, “‘Right-to-Try’ Experimental Drugs: Overview,” 18 23, 2020): 253, https://doi.org/10.1186/s12967-020-02427-4. [xxxvii] Michael S. Okun “Deep Internal Capsule Nucleus Accumbens Region: Responses Observed Active Sham Programming,” Neurology, Neurosurgery Psychiatry 78, 2007): 310–14, https://doi.org/10.1136/jnnp.2006.095315.

Язык: Английский

---

Adaptive Deep Brain Stimulation in Parkinson’s Disease: A Delphi Consensus Study

medRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Авг. 26, 2024

ABSTRACT Importance If history teaches, as cardiac pacing moved from fixed-rate to on-demand delivery in 80s of the last century, there are high probabilities that closed-loop and adaptive approaches will become, next decade, natural evolution conventional Deep Brain Stimulation (cDBS). However, while devices for aDBS already available clinical use, few data on their application technological limitations so far. In such scenario, gathering opinion expertise leading investigators worldwide would boost guide practice research, thus grounding development aDBS. Observations We identified academically experienced DBS clinicians (n=21) discuss challenges related A 5-point Likert scale questionnaire along with a Delphi method was employed. 42 questions were submitted panel, half them being technical aspects other Experts agreed become 10 years. present although panel applications require skilled algorithms need be further optimized manage complex PD symptoms, consensus reached safety its ability provide faster more stable treatment response than cDBS, also tremor-dominant Parkinson’s disease patients those motor fluctuations dyskinesias. Conclusions Relevance Despite concluded is safe, promises maximally effective fluctuation dyskinesias therefore enter into years, research focused markers symptoms.

Язык: Английский

---