Delirium is defined as an acute disturbance in attention, awareness, and cognition that arises as a direct consequence of a physiological event such as an acute illness or infection. It affects roughly one in four older acute medical inpatients, a figure that has remained stubbornly stable across four decades of research. Far from being a benign and transient confusion, delirium is a red flag of underlying vulnerability. It is associated with increased mortality, longer hospital admissions, long-term cognitive decline, poorer functional status, and higher rates of institutionalisation. Crucially, it is also partially preventable: studies suggest that up to a third of new delirium cases in hospitalised patients could be avoided through multicomponent interventions such as reorientation, sleep protection, early mobilisation, and ensuring patients have their glasses and hearing aids. But prevention depends on knowing who is at risk, and that is where the system frequently breaks down.

The Groningen team, led by geriatric medicine researcher Maaike A. Pouw, set out to compare three distinct approaches to delirium identification in the same cohort of patients. The first was a retrospective extraction of documented prehospital indicators: researchers combed through the free-text referral notes and handovers written by general practitioners and ambulance professionals, marking a case as positive when the notes contained an explicit suspicion of delirium or descriptions consistent with it, such as a patient being progressively confused, disoriented, or incoherent. The second was the VMS delirium risk score, a three-question tool used across Dutch hospitals since 2012 that asks about memory problems, need for help with self-care, and any previous episode of delirium. The third was the 4AT, a widely validated four-item screening instrument that assesses alertness, abbreviated mental test performance, attention, and acute mental status change, and can be administered without special training even to drowsy or agitated patients.

The study drew on Acutelines, a hospital-based data and biobank that enrols patients with acute conditions arriving at the emergency department of the University Medical Center Groningen. From September 2020 to February 2024, the researchers included patients aged 70 and older who were admitted for at least one night after their emergency visit and who had a completed 4AT assessment at admission. In total, 655 patients entered the analysis, with a median age of 76 years and 62 percent male. To establish whether patients actually experienced delirium during hospitalisation, the team performed a careful chart-based review using DSM-5 criteria, supplemented by a check for new antipsychotic prescriptions or elevated scores on the Delirium Observation Screening Scale. Uncertain cases were discussed in consensus meetings, a process that applied to 20 percent of all records.

The headline finding was stark. Delirium occurred in 146 of the 655 patients, or 22 percent, consistent with the well-documented burden of the syndrome in older inpatients. When the researchers calculated positive likelihood ratios, a statistical measure of how strongly a positive result points toward the condition, the documented prehospital indicators scored 5.54, the 4AT scored 2.39, and the VMS risk assessment scored just 1.55. In practical terms, when a general practitioner or ambulance medic had written down a suspicion of delirium, that observation was by far the strongest early predictor that the patient would indeed be delirious during the admission. The positive predictive value of the prehospital documentation was 0.64, meaning nearly two-thirds of flagged patients were confirmed cases, and the 4AT achieved an even higher positive predictive value of 0.83, although it flagged only 22 percent of patients as positive.

Each tool, however, told only part of the story. The documented prehospital indicators had a low sensitivity of 0.34, meaning most patients who went on to develop delirium had no such note in their referral documentation. The researchers caution that this may partly reflect documentation gaps rather than a true absence of symptoms at the time of prehospital contact, since emergency crews understandably prioritise physical stabilisation and may not record or convey observations of confusion during a hectic transfer. Among patients with negative prehospital indicators, 16 percent still developed delirium, a figure similar to the 13 percent false-negative rate of the 4AT. The VMS score behaved differently altogether: as a risk-factor-based tool rather than a feature-based one, it generated many false positives, 231 patients or 38 percent of the cohort, but relatively few false negatives, suggesting it is better suited to casting a wide preventive net than to confirming diagnosis.

The study also revealed troubling inconsistencies in routine screening practice. Although compliance with VMS screening was high at 94 percent, exceeding expectations from prior literature, discrepancies appeared in 12 percent of cases. Some patients had been screened twice at different time points with contradictory answers, and in other instances the screening indicated the patient had never experienced delirium even though a previous episode was clearly documented in their medical history. The authors suggest this may reflect the tool’s reliance on patient self-report, since patients may not recall a previous delirium episode, particularly when it occurred on top of underlying dementia. These reliability concerns matter because the VMS score drives decisions about preventive care across Dutch hospitals.

Another striking observation was the limited overlap between the two feature-based approaches. Both the prehospital documentation and the 4AT aim to capture the actual clinical features of delirium at roughly the same point in the care pathway, yet their positive classifications diverged substantially. The authors attribute this to the inherently fluctuating course of delirium, differences in assessment timing, and the possibility that early symptoms resolve quickly after simple interventions such as intravenous fluids or oxygen. They also note that research settings tend to yield higher detection rates than routine care, and that the 4AT in this study was administered partly by trained researchers rather than solely by clinical staff.

The implications reach beyond the emergency department. The authors argue that it would be beneficial to mandate the documentation of cognitive functioning in referral letters and ambulance reports for older patients conveyed to the emergency department, potentially enabling earlier identification and faster intervention, such as involving family members or placing the patient in a calmer environment. They point to emerging technologies that could help, including targeted training for emergency medical services personnel, clinical decision support systems, and natural language processing tools capable of extracting relevant information from clinical narratives automatically. The Institute for Healthcare Improvement recommends delirium screening at least every 12 hours in acute care, but the Groningen team suggests assessment may need to begin even earlier, at the first point of contact at home or in the ambulance.

The study has limitations that the authors acknowledge candidly. It was a single-centre cohort, the medical records were not independently assessed by two reviewers, and the retrospective design meant outcome parameters were restricted to what could be extracted from electronic records. Verbal observations communicated at handover may never have been written down, and chart-based delirium detection depends on the recognition skills of the initial care provider. The interval between early assessment and the occurrence of delirium during a long hospital stay also complicates interpretation, since an initially negative assessment may have been accurate at the time. Nevertheless, the large sample size, multi-year data collection, and realistic representation of routine emergency practice give the findings weight. As the researchers conclude, all three approaches were associated with delirium during hospitalisation, but the humble prehospital note, often scribbled under pressure in the back of an ambulance, may be an underused and valuable asset in protecting older patients from one of medicine’s most consequential syndromes.

Subject of Research: Comparison of three delirium identification approaches in older emergency department patients

Article Title: Comparing three approaches to delirium identification in older patients presenting to the emergency department

Article References: Pouw, M. A., Haan, E. G., Jansen, C. J., Trzpis, M., & van Munster, B. C. (2026). Comparing three approaches to delirium identification in older patients presenting to the emergency department. European Geriatric Medicine. https://doi.org/10.1007/s41999-026-01612-w

Image Credits: AI Generated

DOI: 10.1007/s41999-026-01612-w

Keywords: delirium, emergency department, older patients, 4AT screening, VMS risk score, prehospital care, geriatric medicine, screening tools, likelihood ratio, hospitalisation, cognitive impairment, ambulance notes

Familial searching rests on a simple biological premise. Close relatives share more of their genome than unrelated people do, so a database search that ranks candidates by genetic similarity can point investigators toward family members of the true source of crime scene DNA. The technique gained fame in cases such as the California hunt for the Grim Sleeper serial killer, and it has since become a standard, if controversial, tool in several countries. In the Israeli case, a CODIS-based familial search flagged one male candidate as the highest-ranked potential relative of the unknown offender. The next step was to quantify just how closely related the candidate and the crime scene donor might be, using pairwise kinship analysis based on the autosomal short tandem repeat loci shared between the two profiles.

Short tandem repeats, or STRs, are the repetitive DNA sequences that form the backbone of forensic identification. Each person carries two alleles at each STR locus, one inherited from each parent, and the specific combination of repeat lengths across roughly twenty loci constitutes a DNA profile that is effectively unique to an individual. For kinship questions, however, the analysis is statistical rather than categorical. Analysts compute likelihood ratios, comparing the probability of observing the shared alleles if two people hold a particular relationship, such as parent and offspring, against the probability if they are unrelated. In this case, the analysis over 15 shared autosomal STR loci produced a striking result: the highest likelihood ratio, 530,300, arose under the parent–offspring hypothesis. On its face, the statistics appeared to say that the crime scene donor was the child of the database candidate.

But a parallel line of evidence told a different story. Y-STR analysis, which examines markers on the Y chromosome passed essentially unchanged from father to son, was performed alongside the familial search. It excluded a shared paternal lineage between the candidate and the crime scene donor. That finding was difficult to reconcile with a parent–offspring relationship, at least a father–son one, and it prompted the investigators to look at the other side of the family tree. Full mitochondrial genome sequencing was then carried out, and it demonstrated an identical haplotype between the two profiles, supporting a shared maternal lineage. The combination of an excluded paternal line and an identical maternal line pointed toward a relationship structured very differently from parent and child.

Investigative inquiries resolved the puzzle. The true source of the crime scene DNA was a maternal uncle of the database candidate, an avuncular relationship in forensic terminology. The uncle’s reference STR profile fully matched the evidentiary profile from the burglary scenes, closing the case genetically. What remained open was the scientific question of how a second-degree relationship, uncle to nephew, could generate a likelihood ratio so extreme that it outperformed the parent–offspring hypothesis itself. The answer lies in the stochastic nature of allele inheritance and in the limited number of markers used in routine familial searches.

Parent–offspring relationships are, in principle, the easiest to detect: a child must share exactly one allele with the parent at every locus. Second-degree relatives, such as uncles, nephews, half-siblings, and grandparents with grandchildren, share only about half as much, and the expected genetic overlap is correspondingly lower. Yet chance can intervene. By random luck in meiosis, an uncle and nephew can happen to share alleles at many loci far more often than their relationship predicts, particularly when the marker set is small. With only 15 autosomal STR loci, the tail of the second-degree distribution can reach into territory normally occupied by first-degree relationships, creating overlap between relationship categories that the likelihood ratio framework does not inherently prevent.

To test whether the observed similarity was truly anomalous, the team ran population-based simulations. They generated 100,000 simulated second-degree relative pairs and 100,000 unrelated pairs and examined how the likelihood ratios distributed. The observed second-degree likelihood ratio from the CODIS high-stringency search was 5,538. The simulations showed that 2.05 percent of simulated second-degree pairs produced likelihood ratios equal to or greater than that value. In other words, the striking similarity between the uncle and the crime scene profile, while sitting in the upper tail of the distribution, remained within the range expected for genuine second-degree relatives. The signal was real, but its magnitude was a matter of luck rather than proof of a closer bond.

The most instructive part of the case came in a retrospective expanded-marker analysis. When the investigators increased the number of shared autosomal STR loci from 15 to 20, the parent–offspring hypothesis collapsed entirely, with a likelihood ratio of zero, while hypotheses of close relationships such as second-degree kin remained supported. Five additional markers were enough to completely overturn the ranking that had initially pointed to a parent–child link. This demonstrates a principle with wide implications: relationship inference is not a fixed property of two profiles but a function of how many independent markers are examined. Adding markers increases the statistical power to discriminate between relationship classes and shrinks the zone of ambiguity where stochastic allele sharing can masquerade as a closer kinship.

The authors argue that the case carries a clear lesson for laboratories conducting familial searches. Integrating sufficiently informative autosomal markers, lineage-specific markers such as Y-STRs and mitochondrial DNA, rigorous statistical evaluation, and investigative information is essential when interpreting familial DNA search results. Lineage markers are particularly powerful in this context because they are inherited as single units: a shared Y chromosome implies a shared paternal line, and an identical mitogenome implies a shared maternal line, information that autosomal likelihood ratios cannot provide on their own. Had the Y-STR exclusion been ignored in favor of the eye-catching parent–offspring likelihood ratio, the investigation might have pursued the wrong family configuration entirely.

As familial searching and investigative genetic genealogy expand across jurisdictions, cases like this one illustrate both the power and the fragility of kinship statistics. The technique succeeded here: a database search on a relative’s profile ultimately identified the perpetrator of six burglaries through his maternal uncle. But the path to that identification ran through a misleading statistical signal, an excluded paternal lineage, and a confirming maternal haplotype before the true relationship emerged. For forensic scientists, the message is that no single number, however large its likelihood ratio, should be allowed to dictate a conclusion. Relationship categories overlap at the margins, and only a layered approach combining more markers, lineage evidence, and human investigation can reliably separate an uncle from a father.

Subject of Research: Forensic familial DNA searching and kinship analysis distinguishing avuncular from parent–offspring relationships

Article Title: An avuncular relationship generating a parent–offspring signal in forensic familial DNA searching: a case report

Article References: Schayek, H., Levav-Cohen, Y., Kirshenbaum, L., Rashkovski, K., Avrahami, K., Shenfeld, M., & Starinsky-Elbaz, S. (2026). An avuncular relationship generating a parent–offspring signal in forensic familial DNA searching: a case report. International Journal of Legal Medicine. https://doi.org/10.1007/s00414-026-03987-8

Image Credits: AI Generated

DOI: 10.1007/s00414-026-03987-8

Keywords: familial DNA searching, forensic kinship analysis, autosomal STRs, avuncular relationship, likelihood ratio, lineage markers, Y-STR, mitochondrial DNA, CODIS, short tandem repeats, Israel Police, second-degree relatives

Forensic kinship testing has long relied on short tandem repeats, the repetitive DNA sequences that underpin standard DNA profiling. STRs are excellent for identifying individuals and close relatives such as parents and children, but their statistical power fades rapidly for more distant relationships. As forensic geneticists increasingly turn to sequencing technologies and dense SNP data, a central question has emerged: how many SNPs are actually needed to resolve a given degree of relatedness, and can computational methods extract more information from a moderate number of markers than classical likelihood approaches alone? The new study addresses that question directly, testing a panel of nearly 2,000 SNPs that had previously shown strong forensic value in Chinese Han populations but had not yet been validated in other groups.

The researchers focused their kinship analysis on the Chinese Yunnan Zhuang group, a Tai-Kadai-speaking ethnic minority from Yunnan province in southwestern China. Their goal was to determine whether the moderate-density SNP panel could distinguish first-degree relatives, which include parent-child and full sibling pairs, second-degree relatives such as grandparents, grandchildren, half siblings, and avuncular pairs, third-degree relatives such as first cousins, and unrelated individuals. To do this, they computed two complementary measures for every pair of individuals: the logarithm of the likelihood ratio, a classical forensic statistic that compares the probability of the observed genotypes under a kinship hypothesis versus an unrelated hypothesis, and the cumulative identity-by-state score, which tallies how many alleles two people share at each marker simply by counting matching states.

The results were striking for close relatives. The density curves of the Log10 likelihood ratio and the cumulative identity-by-state values for first- and second-degree kinships were completely separated from those of unrelated individuals, meaning that with these markers, no close relative would be misclassified as unrelated or vice versa. Third-degree kinship proved harder, as expected, because first cousins share on average only about 12.5 percent of their DNA and the overlap with unrelated pairs becomes substantial. When the team set Log10 likelihood ratio thresholds at minus 4 and 4 to separate third-degree relatives from unrelated individuals, the system achieved a power of 86.40 percent with an error rate of zero. In practical terms, the panel can confidently exclude unrelated pairs and correctly flag a large majority of cousin-level relationships without producing false kinship calls, a critical property in forensic and disaster victim identification contexts where a false positive can have serious consequences.

Classical likelihood statistics, however, were only half of the story. The team also trained machine learning classifiers to sort pairs of individuals into four categories: first-, second-, third-degree relatives, and unrelated pairs. Using the Log10 likelihood ratio and cumulative identity-by-state values as input features, they compared several algorithms and found that the K-Nearest Neighbor model performed best, achieving an F1 score of 0.9781 across all four kinship classes. The F1 score, which balances precision and recall on a scale from zero to one, indicates that the model made very few false assignments across the full spectrum of relationship types. This result demonstrates that even a simple, interpretable machine learning method can capture patterns in SNP sharing statistics that fixed thresholds miss, effectively learning the boundary regions where likelihood ratios for different relationship classes overlap.

Beyond kinship, the study tackled a second forensic challenge: biogeographic ancestry inference at fine scale. The researchers integrated genetic data from 135 populations spanning nine geographic regions, drawn from public datasets, and conducted comprehensive population genetic analyses. The Yunnan Zhuang group showed the closest genetic affinity with geographically adjacent populations, particularly the Dai, Miao, and Tujia minorities of southern China and the Kinh population of Vietnam in Southeast Asia. This pattern is consistent with the known population history of the region, in which Tai-Kadai, Hmong-Mien, and related language groups share deep ancestral connections across southern China and mainland Southeast Asia.

Principal component analysis, phylogenetic tree construction, and ADMIXTURE-based ancestry modeling all revealed significant genetic structure among East Asian subpopulations, with especially clear separation between minorities from southern and northern China. Principal component analysis projects individuals into a low-dimensional space defined by the axes of greatest genetic variation, allowing clusters corresponding to population groups to emerge visually. Phylogenetic trees summarize the branching relationships among populations based on genetic distance, while ADMIXTURE models each individual’s genome as a mixture of ancestral components. The fact that all three approaches, applied to a panel of only about 2,000 SNPs, recovered this structure suggests the marker set captures enough ancestry-informative variation to serve as a practical tool for subpopulation discrimination, not just a research-grade dataset.

To push the discrimination further, the team applied multinomial LASSO regression, a feature-selection method that shrinks the coefficients of less informative markers to zero, effectively screening the full panel down to the most ancestry-informative subset. This process yielded 688 SNPs capable of distinguishing five East Asian subpopulations. Using these selected markers, the researchers built two classification models based on different strategies. A partial least squares-discriminant analysis model organized around a hierarchical classification approach, which first splits populations into broad groups and then subdivides them, and an Elastic Net model using a flat multi-classification approach, which assigns each sample directly to one of the five groups, both achieved overall accuracies above 0.9. The convergence of two methodologically different pipelines on similarly high accuracy strengthens the conclusion that the 688-SNP subset carries genuine, robust ancestry signal rather than artifacts of any single algorithm.

The implications for forensic practice are considerable. A moderate-density panel of this size can be genotyped efficiently with targeted sequencing or microarray platforms, at a fraction of the cost and data volume of genome-wide arrays, yet it delivers kinship resolution approaching that of much denser datasets for relationships out to second degree, and usable resolution for third degree when combined with machine learning classification. At the same time, the panel’s ability to distinguish southern from northern Chinese minorities and to place an unknown sample within East Asian substructure could help investigators narrow the geographic origin of unidentified remains or refine investigative leads. The study also adds to a growing literature on machine learning in forensic genetics, where algorithms are increasingly used to squeeze additional inferential power from standard marker sets rather than simply expanding the number of loci typed.

The authors note that the SNP panel had previously demonstrated strong forensic application value in Chinese Han populations, and the present validation in the Yunnan Zhuang group extends its evidence base to another major Chinese ethnic group. The work was funded by the Guangdong Provincial Science and Technology Program and the National Natural Science Foundation of China, and the underlying data are available from the corresponding author upon reasonable request, subject to privacy and ethical restrictions. As forensic laboratories worldwide weigh the transition from STR-centric workflows to SNP-based and sequencing-based platforms, studies like this one provide a practical benchmark: roughly 2,000 well-chosen SNPs, analyzed with a combination of classical likelihood statistics and accessible machine learning classifiers, appear sufficient to resolve close family relationships and to map fine-scale ancestry across one of the most genetically structured regions of the world.

Subject of Research: Forensic kinship identification and East Asian population discrimination using moderate-density SNPs and machine learning

Article Title: Moderate-density SNPs combined with machine learning method driven kinship identification and East Asian subpopulations discrimination analysis

Article References: Wu, X., Liu, Q., Luo, L., Lan, Q., & Zhu, B. (2026). Moderate-density SNPs combined with machine learning method driven kinship identification and East Asian subpopulations discrimination analysis. International Journal of Legal Medicine. https://doi.org/10.1007/s00414-026-04011-9

Image Credits: AI Generated

DOI: 10.1007/s00414-026-04011-9

Keywords: forensic genetics, SNP panel, kinship identification, machine learning, K-Nearest Neighbor, likelihood ratio, ancestry inference, East Asian populations, Yunnan Zhuang, population genetics, LASSO, International Journal of Legal Medicine