In the International Journal of Extreme Manufacturing, Prof. Jinyou Shao at Xi'an Jiaotong University and his co-workers have resolved the challenge of maintaining electrical reliability under repeated bending by bonding a sturdy copper base directly beneath a reservoir of liquid metal alloy. The pairing functions like an autonomous, self-repairing highway: as bending tears microscopic cracks into the solid copper floor, the liquid alloy overhead immediately flows into the fractures to preserve electrical flow, while the rigid copper foundation bears external pressure so the circuit never chokes under compression.

Professor Kibum Kang from the Department of Materials Science and Engineering, in collaboration with UNIST (President Chong-Rae Park), Hanyang University (President Ki-jeong Lee), and TDS Innovation (Co-CEOs Kibum Kang and Sanghyun Mo), has developed a new growth technique that precisely controls the location of nucleation, the stage at which 2D semiconductor crystals first form, using chemical species released from an oxide barrier.

In the International Journal of Extreme Manufacturing, Prof. Yang Li at Zhejiang University and his co-workers developed a dual-sided stencil printing technique that packs high-capacity energy storage into electrode tracks just 80 μm wide by giving a tungsten stencil two distinct physical personalities—a dual-sided "Janus mask" named after the two-faced Roman god.

Cardiovascular medicine is inherently multimodal, but the data available at the clinical decision point are rarely as complete or as synchronized as benchmark evaluations assume. Recent reviews identify incomplete datasets, heterogeneous data characteristics, and missing modalities as persistent barriers to multimodal clinical AI. A correspondence published in Medicine Discovery argues that the central problem is not how to fuse more data, but how to reason when clinically relevant data are incomplete, temporally misaligned, or no longer representative of the patient's current state. The authors distinguish five linked but non-equivalent problems — conventional multimodal fusion, robustness to missing modalities, temporally aware modeling, uncertainty-aware selective prediction, and clinically informed interpretation of why a modality is unavailable — and treat modality availability as an observation state rather than a data defect. They define deferral operationally, separate it from calibration, uncertainty estimation, and distribution-shift detection, propose hybrid deferral rules that combine non-compensatory clinical hard stops with graded model-derived signals, and set out five candidate conditions for an intended-use-specific minimum evaluation suite. The framework is offered as a complement to existing clinical AI guidance (TRIPOD+AI, PROBAST+AI, FUTURE-AI, DECIDE-AI, CONSORT-AI, SPIRIT-AI) rather than a replacement, and as a conceptual proposal requiring prospective evaluation and multi-stakeholder consensus.