. * * window.aa needs to exist and be usable by ANY module's script, * however early it runs - main-nav's script, for one, runs near * the top of

糖心原创

, well before the real search-insights.js * so the download * begins as early as before. Repeat calls share the same downloads, a * failed load is retried on the next call, and the versions / SRI * hashes live only here. * * Algolia Insights is unaffected: window.aa above is still created in * on every page and queues events until search-insights loads. * * window.loadAlgoliaLibs({ priority: 'high' }) (Oct 2026): used by the * /search page, where the first results wait on these libraries. Chrome * gives async scripts low priority, behind images and fonts; this asks * for high instead (fetchpriority). Only the first call for a library * sets its priority - later calls share that download. */ (function (win, doc) { var LIBS = [ { ready: function () { return !!win['algoliasearch/lite']; }, src: 'https://cdn.jsdelivr.net/npm/algoliasearch@5.52.1/dist/lite/builds/browser.umd.js', integrity: 'sha256-I2oYUIvhx1g6l4lTP+yoERbzLIHKzw4AKaBRQPwIeDY=' }, { ready: function () { return !!win.instantsearch; }, src: 'https://cdn.jsdelivr.net/npm/instantsearch.js@4.98.0/dist/instantsearch.production.min.js', integrity: 'sha256-ngmqZk4WBxUw257aMYZayyNBd983INngagv0jI6Ecfs=' } ]; function loadOne(lib, options) { if (lib.ready()) return Promise.resolve(); // One download per library, shared by every caller; reset on // failure so the next call retries just the library that failed. if (lib.pending) return lib.pending; lib.pending = new Promise(function (resolve, reject) { var script = doc.createElement('script'); script.src = lib.src; script.integrity = lib.integrity; script.crossOrigin = 'anonymous'; script.async = true; if (options && options.priority === 'high') script.setAttribute('fetchpriority', 'high'); script.onload = function () { lib.ready() ? resolve() : reject(new Error('Loaded but not available: ' + lib.src)); }; script.onerror = function () { script.remove(); reject(new Error('Could not load ' + lib.src)); }; doc.head.appendChild(script); }); lib.pending.catch(function () { lib.pending = null; }); return lib.pending; } win.loadAlgoliaLibs = function (options) { return Promise.all(LIBS.map(function (lib) { return loadOne(lib, options); })); }; })(window, document); . */ window.hsConversationsSettings = window.hsConversationsSettings || {}; window.hsConversationsSettings.loadImmediately = false; (function (win) { var EVENTS = ['pointerdown', 'keydown', 'touchstart', 'scroll', 'mousemove', 'wheel']; var done = false; /* 19 Oct 2026 - chat widget layout shift (field CLS). Phones fail Core Web Vitals on CLS (p75 ~0.2 site-wide in CrUX for months). Biggest cause found: HubSpot's chat container is fixed to the bottom-right corner and changes size - 100x96 launcher to 282x253 when the welcome message pops up (~5 s after load), and to full screen when a visitor opens the chat. Its top-left corner moves, so Chrome scores it as a layout shift: 0.088 per page view at 360x640 (0.044 at 412x850), on every page. Taps happen inside the chat iframe, so the parent page never sees the input and the open/close shifts count too. Fix: whenever the container's size changes it is made transparent for 400 ms (Chrome doesn't score shifts of invisible elements), then shown again at its new size. Tested on live io1001 at 360x640 in a cookie-less frame: the 0.088 shift no longer appears. The chat itself is unchanged - it just appears 0.4 s later at each resize. Relies on HubSpot's container id; if that changes this does nothing (no error). */ function cloakChatResizes() { var ID = 'hubspot-messages-iframe-container'; function watch(c) { if (c.getAttribute('data-bb-cloak')) return; c.setAttribute('data-bb-cloak', '1'); var last = c.offsetWidth + 'x' + c.offsetHeight, timer = null; new MutationObserver(function () { var size = c.offsetWidth + 'x' + c.offsetHeight; if (size === last) return; last = size; c.style.setProperty('opacity', '0', 'important'); clearTimeout(timer); timer = setTimeout(function () { c.style.removeProperty('opacity'); }, 400); }).observe(c, { attributes: true, attributeFilter: ['style', 'class'], subtree: true, childList: true }); } var found = document.getElementById(ID); if (found) { watch(found); return; } if (!document.body || !('MutationObserver' in win)) return; var mo = new MutationObserver(function () { var c = document.getElementById(ID); if (c) { mo.disconnect(); watch(c); } }); mo.observe(document.body, { childList: true }); setTimeout(function () { mo.disconnect(); }, 30000); } function loadWidget() { var api = win.HubSpotConversations; if (api && api.widget && api.widget.status && api.widget.status().loaded) return; try { cloakChatResizes(); } catch (e) {} if (api && api.widget) api.widget.load(); } function onFirstInteraction() { if (done) return; done = true; EVENTS.forEach(function (type) { win.removeEventListener(type, onFirstInteraction, { passive: true }); }); var start = function () { if (win.HubSpotConversations) loadWidget(); else (win.hsConversationsOnReady = win.hsConversationsOnReady || []).push(loadWidget); }; if ('requestIdleCallback' in win) win.requestIdleCallback(start, { timeout: 1500 }); else setTimeout(start, 200); } EVENTS.forEach(function (type) { win.addEventListener(type, onFirstInteraction, { passive: true }); }); })(window);

糖心原创

Skip to main content
Blog
Neuroscience 2025: The shift toward integrated human models

19.12.2025 | Published by 糖心原创

Neuroscience 2025: The shift toward integrated human models

Featured Image

The momentum at this year's Society for Neuroscience meeting was unmistakable. Across posters, talks, and discussions, the narrative around human in vitro models has shifted. The focus is no longer on on whether and how we can generate functional human cells from iPSCs, but on how to build models that are mechanistically informative, multicellular, and predictive. Researchers are moving beyond simple monocultures toward integrated systems that capture circuit-level behaviour and cell鈥揷ell interactions. SfN 2025 made it clear: the field is now thinking in terms of function, complexity, and translational relevance.

Multi-regional brain models are becoming the new standard

One of the clearest signs of this shift is the rise of multi-regional brain models, or assembloids. By combining distinct organoids (cortical, thalamic, hippocampal) researchers can study how brain regions connect and communicate, giving access to circuitry that drives human behaviour.

Assembloids vs standard organoids: What is the difference?

Standard organoids | Region-specific identity; limited connectivity

Assembloids | Inter-regional communication, functional circuity, network phenotypes

Assembloids provide access to the circuit-level mechanisms that underlie many neurodevelopmental and neuropsychiatric disorders

 

A standout example came from Rebeca Blanch (, University of California San Diego), who compared cortico-thalamic assembloids with standard organoids in modelling CDKL5 Deficiency Disorder. The fused models captured the E/I imbalance and maturation defects associated with epilepsy in a functionally integrated, circuit-level model.

This is a significant step forward for modelling neurodevelopmental disorders such as autism and schizophrenia, where pathology is rooted in connectivity rather than isolated cellular deficits.

Engineering meets biology: toward scalable, patient

While assembloids address anatomical complexity, engineering-led innovation is tackling scale and physiological relevance. A dominant theme was the patient-on-a-chip approach, integrating microfluidics, high-content imaging, and iPSC biology. These platforms enable high-throughput testing of compounds on cells carrying a patient鈥檚 genetic background, an operationalisation of personalised medicine.
  
Melissa Hernandez (Salk Institute) showcased a compelling transdifferentiation workflow that scales 2D neurons and astrocytes into 3D constructs while retaining the donor鈥檚 ageing signature, which is often erased during reprogramming.

Coupled with advanced MEA systems capturing real-time network dynamics, these models open the door to functional screening for disorders such as 础濒锄丑别颈尘别谤鈥檚 and ALS that demand network-level readouts.

If you are exploring functional activity in human neuronal systems, you can find MEA datasets and protocols on the 糖心原创 website. 

Glia take centre stage in next-generation disease models

SfN 2025 made clear that advancing disease relevance requires moving beyond neurons alone The role of glia, astrocytes, microglia, and oligodendrocytes, was central across the conference. 

Researchers presented increasingly complex co-culture and tri-culture systems to model: 

  • Microglial-driven neuroinflammation in Parkinson's disease
  • Oligodendrocyte-mediated myelination deficits in multiple sclerosis
  • Astrocyte-neuron interactions that drive neurodegeneration

These systems expose therapeutic targets that remain invisible in neuron-only cultures, underscoring the need for integrated human models.

To see how defined human iPSC-derived glial cells are used in co-culture and mechanistic studies, visit the glia product and application pages on the 糖心原创 site

 鈥淔rom the perspective of someone deeply involved in functional assay development, the challenge is no longer biological feasibility, but operationalisation. Moving from robust 2D systems to complex 3D models demands improvements in reproducibility, scale, and standardisation. Academic-industry collaboration will be key, especially for MEA interpretation and functional readouts鈥�

Luke Foulser
Luke FoulserField Applications Scientist
糖心原创
 
 
 
 
 

The direction is clear, and the field is accelerating

Despite the challenges, the trajectory is encouraging. Advances in cell programming, chip technologies, and AI-driven imaging are removing barriers that previously slowed adoption of complex human models. 

SfN 2025 left one message resonating through the community; we are entering a phase where human in vitro models are no linger proofs of concept, they are becoming functional systems for discovery.

For practical resources, including culturing and assay protocols, tutorial videos and troubleshooting tips, you can access the 糖心原创 support hub.

 

FAQs

What were the major trends in human in vitro neuroscience models at SfN 2025? 

The major trends in human in vitro neuroscience models at SfN 2025 centred in the shift toward integrated, mechanistically informative systems. Researchers moved beyond simple monocultures toward multi-regional brain models, patient-specific microengineered platforms, and glia-inclusice co-cultures that capture circuit-level behaviour and disease mechanisms more accurately than previous models


Why are assembloids important for disease modelling? 

Assembloids are important for disease modelling because assembloids recreate inter-regional connectivity, allowing researchers to study disorders rooted in circuit dysfunction such as CDKL5 deficiency disorder, autism, schizophrenia. By fusing cortical, thalamic, or hippocampal organoids, assembloids provide access to network-level phenotypes that cannot be reproduced in isolated region-specific organoids. 


How are MEA systems used in disease models? 

MEA systems are used in disease models because MEA platforms quantify network activity, synchrony, and firing patterns, providing functional readouts that reflect circuit-level phenotypes. This makes MEA essential for studying diseases where network behaviour is a defining feature - including epilepsy, Alzheimer's disease, ALS, and other disorders where synaptic or circuit dysfunction emerges before molecular endpoints. 


Related pages

About us Coding cells to advance the wellbeing of humanity
Careers See the latest openings
News See the latest news