Arslan, Adem; Gül, Fatih

Background – Early identification of neurodevelopmental conditions during the pre-school years (ages 3–6) is critical for timely intervention, yet traditional questionnaire-based screeners suffer from low positive predictive value, late detection, and inequitable access. Artificial intelligence (AI) and machine-learning (ML) methods promise more objective, scalable identification, and two devices (Cognoa Canvas Dx, 2021; EarliPoint Evaluation, 2023) are already US Food and Drug Administration (FDA)-authorized. Recent reviews summarize the technical literature, but few address how deployed tools perform on clinical validity, equity, and real-world implementation, particularly outside high-income, English-speaking settings. Methods – Following Joanna Briggs Institute methodology and the PRISMA extension for scoping reviews (PRISMA-ScR), we searched seven databases (1 January 2020–15 June 2026) and charted primary validation studies of AI/ML developmental-screening tools for children aged 72 months or younger. We mapped tools by modality, diagnostic performance, regulatory status, geographic origin, and, as a structured analytic axis, demographic subgroup (equity) reporting and deployment readiness. Results – Sixteen primary AI-tool studies met inclusion, spanning eight modality classes (eye-tracking, multimodal questionnaire-plus-video, computer-vision home/clinical video, tablet/smartphone digital phenotyping, voice/acoustic ML, electronic-health-record risk prediction, questionnaire-based ML, and wearable sensors) and eight countries (USA, South Korea, Australia, Switzerland, France, Israel, China, Vietnam). Reported sensitivities ranged from approximately 75 to 99.2% and specificities from 78.9 to 95.5%. Only one tool demonstrated stable performance across sex, race, and ethnicity; subgroup-stratified validation was otherwise near-absent. No CE-marked autism-specific screening device was identified. Conclusions – The evidence base has broadened in modality and geography but remains autism-concentrated and equity-blind, with a wide gap between analytic accuracy and primary-care deployment. We propose equity validation, deployment-readiness reporting, and an implementation pathway for low-resource guidance-center settings as priorities.

Gökgöz, Baki; Aydin, Tolga; Gül, Fatih

The traditional Von Neumann computing architecture, which necessitates data transfer between external memory and the processor, incurs significant energy and time costs when running deep learning (DL) and machine learning (ML) architectures. The primary issue with the energy and time efficiency of this architecture stems from the frequent and intensive data transfers between memory and the processor. Therefore, memristive synaptic devices are utilized to overcome this energy and time inefficiency while performing cognitive tasks. The fundamental working principle of memristive devices is to reduce the need for data transfer by combining memory and processing in the same location, thereby significantly decreasing both energy consumption and the time required for operations. However, to achieve the desired level of efficiency in terms of energy and time consumption from neuromorphic systems, the performance of these systems needs to be further improved with respect to accuracy and test error rates for classification applications. Achieving high accuracy performance in such deep learning or machine learning models necessitates optimization processes not only at the hardware level but also at the algorithmic level. In this context, this paper presents a comprehensive examination and comparison of the frequently used SGD and its momentum variants for deep learning and machine learning applications in memristor-based neuromorphic computing systems. The study thoroughly investigates the performance of critical metrics such as the learning properties, energy efficiency, and accuracy rates of the nano-scale titanium dioxide (TiO2) based synaptic device. The experimental results for the MNIST dataset showed AdaDelta 89.48%, AdaGrad 79.00%, Adam 79.13%, AdaMax 79.68%, Momentum 88.55%, Nadam 81.20%, RMSprop 84.91% and SGD 89.47% accuracy. The experimental results for the CIFAR dataset showed AdaDelta 90.51%, AdaGrad 82.08%, Adam 83.10%, AdaMax 81.76%, Momentum 91.25%, Nadam 82.45%, RMSprop 88.11% and SGD 90.21% accuracy. 2024 The Authors. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.

Gökgöz, Baki; Gül, Fatih; Aydın, Tolga

The prevalence of artificial intelligence applications using artificial neural network architectures for functions such as natural language processing, text prediction, object detection, speech, and image recognition has significantly increased in today's world. The computational functions performed by artificial neural networks in classical applications require intensive and large-scale data movement between memory and processing units. Various software and hardware efforts are being made to perform these operations more efficiently. Despite these efforts, latency in data traffic and the substantial amount of energy consumed in data processing emerge as bottleneck disadvantages of the Von Neumann architecture. To overcome this bottleneck problem, it is necessary to develop hardware units specific to artificial intelligence applications. For this purpose, neuro-inspired computing chips are believed to provide an effective approach by designing and integrating a set of features inspired by neurobiological systems at the hardware level to address the problems arising in artificial intelligence applications. The most notable among these approaches is memristor-based neuromorphic computing systems. Memristors are seen as promising devices for hardware-level improvement in terms of speed and energy because they possess non-volatile memory and exhibit analog behavior. They enable effective storage and processing of synaptic weights, offering solutions for hardware-level development. Taking into account these advantages of memristors, this study examines the research conducted on artificial neural networks and hardware that can directly perform deep learning functions and mimic the biological brain, which is different from classical systems in today's context.

Gül, Fatih

Resistive random access memories (RRAMs) are favorable contenders in the race towards future technologies. Moreover, the desirable properties of memristor-based RRAM devices make them very good competitors in this field. The sneak paths problem poses one of the main difficulties in the construction of crossbar memory devices. This problem can be effectively suppressed by applying the 1diode-1resistor (1D1R) design structure. The Schottky diode has many advantages compared to the PN junction diode. The low-power (∼1 µW) ZnO active-layered thin-film (60 nm thick) one Schottky diode-one memristor device fabricated in this study included a top Ag electrode and a bottom Al electrode. The material makeup of the device was confirmed via energy dispersive X-ray spectroscopy (EDAX). The memristive and Schottky diode characteristics of the Ag/ZnO/Al device were resolved by measuring the time-dependent voltage/current. The characteristic pinched hysteresis memristive loops were observed at the first quadrant of the current-voltage plane, whereas the diode curves were seen at the third quadrant. Using the current-voltage curves, the height of the Schottky barrier, ideality factor and threshold voltage of the Schottky diode were found to be 0.68 eV, 3.75 and 0.49 V, respectively. After physical implementation and characterization of the one diode-one memristor device, its anti-crosstalk characteristics were investigated. Taking into account the 10% read margin, the maximum crossbar size was found to be 87.