Emerging Novel SARS‐CoV‐2 Subvariants and the Advantages of AI‐ML in Deciphering the Mutation Trends, Genomic Surveillance and Vaccine Development

ABSTRACT

Background and Aim

The world witnessed COVID‐19 in 2025 yet again, a resurgence driven by LF.7 and NB.1.8.1 subvariants of JN.1. With Singapore, Hong Kong and Thailand as the epicenter this time, these variants demonstrated rapid geographic spread and temporal dominance across regions including India, the United States (US), the United Kingdom (UK), and China. Genomic surveillance highlighted a sharp peak due to JN.1, followed by rising trends due to LF.7 and NB.1.8.1. Thus, a perspective was assumed aiming to get insights into the shifting epidemiological and transmission dynamics of COVID‐19, and how recent advanced technologies like AI‐ML could play a key role in deciphering the mutation trends of the novel virus, global genomic surveillance and vaccine research and development.

Methods

Literature on recent global COVID cases was sourced online using reliable dedicated search engines. The limited available data from sources was used to get insights. Data were further analyzed, primarily stressing on the molecular‐level interplay within the virus and the host, possible technological interventions to know the infectivity and predict the transmission patterns, and the role of artificial intelligence and machine learning integration as solutions for vaccine designing to strategize community health.

Results

India recorded over 3900 active cases in the year 2025 till early June. Kerala and Maharashtra contributed the majority of the fresh cases. Most reported cases were clinically mild, but isolated severe outcomes were witnessed, particularly among the high‐risk subjects. Despite successful mass vaccination drives, recent transmission suggested a dynamic interplay of the waning immunity, adaptability of the subvariants, and altered public behavioral patterns.

Conclusions

Integrated AI‐ML tools greatly enhanced the real‐time response and resource planning through genome surveillance, predictive modeling, and outbreak predictions. In silico cutting‐edge technologies are useful tools facilitating rapid assessment of the spread and strengthening decision‐making in public healthcare. Designing AI‐assisted strategies and continued monitoring were vital to mitigate threats of infectious diseases in the future.

Summary

• The 2025 COVID‐19 resurgence, driven by JN.1 subvariants LF.7 and NB.1.8.1, spatiotemporally expanded quickly from Southeast Asia to many global locations.

• Genomic surveillance revealed synchronized transmission peaks in multiple countries, highlighting the importance of tracking the on‐going rapid mutations.

• Integrated AI‐ML frameworks enable real‐time genome analysis, predicting outbreaks, and improved assessment of the transmission dynamics.

• AI‐assisted in silico platforms significantly accelerated vaccine development by predicting antigens and through immunogenic modeling.

• Strengthening public health resilience to manage pandemic risks through AI‐driven surveillance and translational policy planning was essential.

1. Introduction

The world experienced a sudden resurgence in COVID‐19 cases yet again in 2025 after nearly 3 years of epidemiological stability attributable to vaccination drives [1]. The recent test positivity rate reportedly reached 11% for the first time since July 2024 [1]. The scenario was despite high herd immunity due to active immunity during infection and/or vaccine drives. Recent surveillance noted several circulating SARS‐CoV‐2 subvariants LF.7, XFG, JN.1, and NB.1.8.1 with high transmissibility and mild clinical presentations. Sore throat, fatigue, low‐grade fever, nasal congestion, and gastrointestinal disorder were usually seen. Although the subvariants were classified as low‐risk, the emergence highlighted the progressive virus evolution and waning immunity issues and calls for sustained genome surveillance, timely booster updates, and data‐driven public healthcare responses to contain future occurrences and support global healthcare infrastructure. Although not alarming yet, the uptick came at a time when two important countries in Southeast Asia, Singapore and Hong Kong, battled the fresh COVID‐19 waves. Singapore saw 28% rise in weekly cases, while Hong Kong reported 31 COVID‐related deaths, which was its highest in a year. COVID‐19 cases were globally rising then in the Eastern Mediterranean, South‐East Asia, and Western Pacific countries [1].

India's active COVID‐19 caseload steadily rose to 3961 by June 2, 2025, chiefly in clusters. With over 1400 active cases, Kerala Province led the national tally, followed by Maharashtra (506), West Bengal (287), and the capital region of Delhi (104). COVID‐19‐associated fatalities were simultaneously reported from Kerala and Karnataka in patients with serious underlying comorbidities, highlighting the high‐risk in clinically vulnerable population. Four recent fatalities in India, including that of a 40‐year‐old patient with acute respiratory distress, highlight that emerging subvariants were serious threats to the vulnerable [2]. Most infections remained clinically mild and manageable through quarantine at home. However, emerging cases in Gurugram, Punjab, and Odisha in India hinted at a broader geographic spread of the subvariants. The Union Ministry of Health assured the Indian public of the ongoing surveillance and urged preparedness. Rising cases highlight that real‐time genetic tracking, alertness of the public at large, preparedness of the healthcare professionals, and protection of high‐risk groups were necessary to curb the transmission and prevent future outbreaks [3]. Although most COVID‐19 cases in India and elsewhere were allegedly largely mild, isolated incidents of severe disease and mortality were still being reported.

Efforts at AI‐ML and pandemic management interface have notably increased between 2020 and 2025, with more than 3000 peer‐reviewed articles published worldwide. The US, China, and the UK have become the major contributors through significant research collaborations and continuing international cooperation. Recurrently emerging SARS‐CoV‐2 subvariants highlight the limitations of conventional epidemiology and the need to integrate AI and ML. Extensively cited high‐impact research and reviews on AI‐guided genomic surveillance, epidemic modeling and vaccines advances highlight their implications in preparing for pandemics in the new world order. Despite advantages in analyzing the genome, detecting mutations and mRNA vaccine development, use of AI‐ML is still limited. This perspective expands the potential of AI‐ML to monitor and predict future outbreaks, over and above vaccine development, with regards to upcoming subvariants.

2. Methodology

A comprehensive literature search was conducted to source recent data on the subvariants, the genomic surveillance, and the application of AI/ML in predicting COVID‐19 and developing vaccines. Recent literature published in English between January 2020 and June 2025, was searched by using the terms “COVID‐19 subvariants”, “genomic surveillance”, “AI outbreak prediction”, and “AI in vaccine development”, etc. from reliable databases (PubMed, Scopus, Web of Science). The primary stress was on the molecular level interplay within the viral entity and the host, the possible suitable technological interventions to understand infectivity and predict transmission patterns, and the role of AI‐ML in finding solutions from designing the vaccine to devising community health strategies [4]. Real‐time epidemiological data were also included from the global and national public health and genomic surveillance databases of the World Health Organization (WHO) and Centres for Disease Control and Prevention (CDC). We got 225 articles from the initial screening. After the full‐text review and following the inclusion and exclusion criteria, a total of 39 articles with specific data and comprehensive discussion were included in the final analysis for a detailed narrative synthesis.

3. Emerging Variants and Their Molecular Bases

Driven primarily by the emerging LF.7 and NB.1.8.1 JN.1 subvariants, recent genomic surveillance data indicated a rise in COVID‐19 cases globally. Distinct mutational patterns within the spike protein (i.e., critical in viral entry and immune recognition) were revealed in NB.1.8.1, LF.7, JN.1, and XFG through genomic analyses. Spike protein in NB.1.8.1 lineage (currently under monitoring by the WHO) had mutated F486P and R346T, both associated with enhanced immune escape and reduced neutralization by monoclonal antibodies [5]. Another subvariant of concern, LF.7 carried an altered receptor‐binding domain (RBD) that could impact the binding affinity and antigenic structure of angiotensin‐converting enzyme inhibitor 2 (ACE2), and detailed structural data were still emerging [6]. JN.1, a BA.2.86 lineage descendant, featured multiple spike substitutes like S486P, N460K, and E180V that are implicated in increased transmissibility and partial resistance to vaccine‐induced immunity [7]. While still under limited characterization, the XFG variant shared some convergent mutations observed in other immune‐evasive Omicron sublineages [8]. These variants collectively displayed a convergent evolution trend, favoring mutations that enhanced receptor interactions and reduced antibody binding. Although clinical severity was low, the accumulation of the adaptive mutations highlighted pressure on the virus to evolve in response to widespread host immunological resistance. Emerging subvariants NB.1.8.1 and LF.7 exhibit convergent evolution characterized by mutations including F486P, Q498R, R346T, P681H, and N679K that are associated with enhanced ACE2 binding and partial escape from neutralizing antibodies. These molecular adaptations contribute to the continued evolution and persistence of immune‐evasive SARS‐CoV‐2 lineages [9, 10].

Country‐wise trend analyses after 2023 revealed a parallel surge due to SARS‐CoV‐2 variants in India, US, and Brazil, while Omicron and its descendants dominated earlier. JN.1 widely persisted in 2024 across all regions, while a marked surge in 2025 was attributed to LF.7 and NB.1.8.1. The area charts in Figure 1 represent variants‐specific infections between 2020 and early 2025, particularly in the three major grossly‐affected nations, the US, Brazil, and India, suggesting a simultaneously‐occurring synchronized infection waves. Past variants (like Delta and Alpha) were rapidly displaced, reflecting the adaptive advantage of the mutated spike in the novel strains that enhanced the transmissibility and provided an immune escape advantage, at least partially. The patterns confirmed that while most nations had achieved high vaccination coverage, waning immunity and the variant‐driven antigenic shifts facilitated global viral transmission and sustenance of the infection. The trends highlighted the importance of a real‐time genome tracking system and updated vaccine formulations to match the circulating variants [11].

Figure 1.

Area charts of COVID‐19 infections by variants between 2020 and 2025, suggesting multi‐location simultaneous synchronized infection waves. The contribution of each variant to the weekly case totals in each area is shown in varying‐colored lines. Emerging overlapping variant and their temporal worldwide transmission depict synchronized multi‐location infection wave phenomena.

While the WHO classifies subvariants LF.7 and NB.1.8.1 as “variants under monitoring” (VUM) currently, their rapid spread in India, China, Singapore, and Hong Kong underscores their high transmissibility. Genetic data of INSACOG confirmed the dominance of NB.1.8.1 in Tamil Nadu and LF.7 in Gujarat, indicating local transmission. Although the infection was generally mild, the few deaths that occurred signaled the need to be vigilant. Mutated spike protein in the variants enhanced the binding property or neutralizing antibodies evasion in the virus, although detailed mutational profiles were still being investigated. The clustering of recent cases in Kerala, Maharashtra, and elsewhere stresses the need for continued genomic surveillance and adherence to preventive measures (https://www.livemint.com/science/health/covid19-cases-rise-in-india-bengaluru-gurugram-report-new-infections-delhi-cm-says-not-alarming-7-updates-11748343212033.html, accessed on May 29, 2025).

4. Analyzing the Reasons Behind the Current Resurgance

Constantly mutating SARS‐CoV‐2 virus makes the existing vaccines ineffective. Also, the acquired immunity wanes after some days of vaccination, which could critically contribute to the resurging cases. Many studies have demonstrated that neutralizing antibody titers, especially against the emerging variants that mutated to escape the immune system, declined significantly in 4–6 months after vaccination. Compromised immunity reduced the vaccine potency to protect not only against the infection but also on the outcomes of symptomatic and severe illnesses [12]. Memory B‐cell and T‐cell responses persisted longer but could not fully prevent viral replication or transmission owing to antigenic drift. Breakthrough infections become more frequent over time as a result, necessitating updated booster strategies tailored for the circulating mutated subvariants [13]. Waning immunity, antigenic drift, altered COVID‐appropriate behavior, viral recombination, and climatic influences could be the confluence of factors attributing to the resurgence of COVID‐19.

Antigenic drift in SARS‐CoV‐2 refers to the gradual accumulation of mutations, particularly in the RBD of the spike protein that altered the antigenic properties. Such mutations reduced the binding efficiency of the neutralizing antibodies generated due to active or passive infection (vaccination) earlier, thereby facilitating immune escape. JN.1, LF.7, and NB.1.8.1 variants exhibited substitutions like F486P, R346T, and Q498R, which significantly diminished antibody neutralization [14]. Ongoing antigenic evolution compromises the efficacy of the existing immunological memory and underscores the need for continuous genomic surveillance and formulation of tailor‐made vaccines [15].

Recombination in SARS‐CoV‐2 occurs in an infected host when it is coinfected with distinct viral lineages, allowing the RNA‐dependent RNA polymerase (RdRp) to switch templates during replication. Hybrid genomes are generated through this by combining advantageous mutations from multiple variants. Notable recombinants like XD, XE, and XBF possessing spike proteins with enhanced immune evasion arose from parental lineages Delta and Omicron through similar mechanisms [16]. Such events, when they occur, accelerate the evolution of virus and potentially lead to high transmissibility, enhanced infectivity, altered antigenicity, or antibody resistance. It emphasizes on robust genomic surveillance in high transmission periods [17] and consistent efforts to update the vaccines.

Climate change influences the transmission dynamics of, especially respiratory viruses like SARS‐CoV‐2, in altered temperature, humidity, and seasonal variation patterns. Fluctuating ambient conditions could affect viral stability, host susceptibility, and human social behavior (like indoor crowding in extreme weather). Studies show that cooler temperatures and lower humidity enhanced the stability of aerosols and survival of the virus, potentially increasing the transmission rates [18]. Moreover, climate‐driven ecological disruptions and urban pollution could weaken host immunity, which would have a compounding infection risk. As SARS‐CoV‐2 adapts to varied environments, climate‐sensitive surveillance systems, particularly at the regional level, are essential to anticipate outbreak patterns.

5. AI‐ML Based Surveillance, Prediction, and Response

The rise of JN.1 variant of SARS‐CoV‐2 (an Omicron BA.2.86 variant) and its subvariants LF.7 and NB.1.8.1 raised critical concerns. The global concerns are largely attributed to the high transmissibility and potential to partially evade vaccine‐induced immunity (Figure 2). Recognized eminently in 2023, JN.1 became the most prevailing strain rapidly in some countries, raising the caution bar, especially in the regions with limited healthcare settings, apparently due to its potential effect on the various wellbeing parameters [19]. JN.1 itself carried a special combination of genetic mutations, the prominent of which was the L455S alteration inside its RBD that contributed to its high transmissibility and reduced antibody neutralization [19].

Figure 2.

Country‐wise JN.1 variant proportions highlighted steep peaks in global circulation during late 2023 and early 2025; comparative rise, peak, and decline of JN.1 stress marked a surge. Country‐specific trends demonstrate a near‐coincident global rise and spread of JN.1.

Despite concerns, available information suggested that JN.1 spread faster, although it did not cause extreme infectivity as compared to the earlier variants [20]. In all possibilities, nevertheless, its quick spread has put healthcare settings under great stress, especially in populated urban areas where human–human infection risk and reduced human productivity were high. Vaccination drive remains the basic and the most effective tool. In the face of such circumstances, the viability of the available immunization programs against JN.1 (or its subvariants) has been a subject of detailed investigation. Whereas antibodies offer security and safety against highly infectious COVID‐19 forms, their adequacy in containing JN.1 contamination is a great concern [7, 21]. It drove scientific deliberations apparently to upgrade vaccines and immunization programmes to contain the circulating variants that were superior on transmissibility (and possibly infectivity too).

WHO prescribed overhauling the COVID‐19 antibodies to target the currently prevailing strains, particularly those that were related to JN.1 [22]. Human wellbeing service agencies in India have expressed that there were no immediate requirements of extra booster dose for the general public, although high‐risk groups were recommended to volunteer for a preventive third dosage. Genomic reconnaissance plays a crucial role, as is understood, in multiple waves of the advent of SARS‐CoV‐2 variants and their spread. Indian SARS‐CoV‐2 Genomics Consortium (INSACOG) was instrumental in sequencing and examining emerging JN.1 variants. AI and ML tools have been progressively vital, especially within the gamut of health emergencies like COVID‐19.

Current AI/ML applications in COVID‐19 management are primarily focused on genomic surveillance, lineage classification, outbreak modeling, and diagnostic support. Physiological monitoring through wearable devices and AI‐assisted vaccine optimization remain as other applications, including in the areas of active research and development. Bioinformatics‐based (in silico) machine learning (ML) and deep learning (DL) AI tools like DeepVariant (variant calling) and Pangolin‐ML (for SARS‐CoV‐2 lineage assignment) did enable and upgrade the exactness of the variants' discovery and classification [23]. These models rely on training data sets wherein the low‐ and middle‐income population might be often underrepresented due to the skeletal surveillance system, leading to predictive bias and potential inequities in public health utility. Studies indicated that ML forecasting using recurrent neural networks, ensemble learning, hybrid ML‐mechanistic models, etc., often outperformed the conventional SEIR (susceptible exposed infectious recovered) models in adaptive, short‐term outbreak prediction under the policy shifts or nonlinear viral dynamics conditionality. However, SEIR frameworks provided superior interpretability and long‐term planning, highlighting the contextual suitability of distinct modeling approaches and the importance of integrating both perspectives in genomic‐driven public health interventions [24].

Platforms like Nextstrain facilitate real‐time phylogenetic analyses, viral evolution visualization, genomic surveillance support, and tracking of variants as they emerge [25]. Wearable gadgets with sensors could collect physiological data like the cardiac and respiratory rates, which are analyzed by AI models to anticipate the onset of possible COVID‐19‐associated effects. AI‐assisted CT imaging technique has improved the ability to diagnose COVID‐19 better, facilitating the convenient affirmation and isolation of seemingly contaminated subjects [26]. Determination of episodes has moreover profited from AI‐driven tools and techniques. Platforms like BlueDot and EpiCast utilize AI and big‐data analytics to predict the spread of an infectious disease based on many parameters, counting on the portability data and social intelligence. These models have been instrumental in educating the general public about the wellbeing preferences, like volunteering for quarantine, lockdowns, COVID‐appropriate behavior, and effective allocation of biomedical assets. However, challenges continue to haunt in spite of these innovative headways. AI models greatly depend on honest reporting of the cases by the general public and other stakeholders, which may not otherwise augur well as statistical cohorts, driving to potential inclinations (predictive bias) while drawing inferences [27]. AI‐assisted computational approaches encompass aspects of vaccine research as in the mRNA vaccine pipelines of Moderna and BioNTech, that include the selection of antigen and optimization of immunogen. Nonetheless, vaccine development relies heavily on experimental validation, preclinical testing, and clinical evaluation. It highlights the transformative role of AI‐ML in epitope mapping to select candidates, optimize codon, predict the stability, genomic surveillance, along with outbreak forecasting and vaccine innovation, thus strengthening the pandemic response efforts [28]. However, equitable implementation of data governance, technical constraints, and biases in disadvantaged groups are still barriers [27].

Also, employing AI tools and techniques in open‐ended wellbeing frameworks needs a strong computational foundation and trained staff to handle big data. Moral issues like information security and education must be oriented for guaranteed capacity utility of AI in healthcare. Resurgence of JN.1 variants underscores needs for flexible techniques in wider surveillance, monitoring, and diligent prediction of COVID‐19. Although immunization drive remains as a safe and fundamental countermeasure option, integrating AI and ML as observing, locating and predicting tools offers promising headways to open and flexible wellness readiness. Interdisciplinary and multidisciplinary investigations and collaborations involving virology, community healthcare innovation, and overall wellbeing could help explore current and future challenges of SARS‐CoV‐2 and its advancing variants (like VOIs and VUMs).

6. AI‐ML as a Tool in Vaccine Development

Global COVID‐19 resurgence has reignited human wellbeing and immunization adequacy concerns on prevailing strains. High transmissibility of JN.1 with potential vaccine resistance risk emphasizes the need for a progressively advanced immunization strategy. AI and ML are promising and essential tools in modern science and technology to predict and determine expected viral mutations, optimizing antigen determinants, hastening vaccine development, and designing foolproof immunization plans. While these approaches show considerable benefits in accelerating early‐stage vaccine research, their predictive performance depends on the quality of data, the designed model, and experimental validation. Designing universal immunization drives, recommending booster doses in time, and resource allocation through equitable distribution could be optimized through AI‐ML tools. AI‐ML‐based techniques have bolstered progress across vaccine development platforms. These platforms could potentially provide broad‐spectrum immunity against multiple SARS‐CoV‐2 strains, including future variants. AI tools analyze vast data sets to identify conserved epitopes across various strains, providing critical outputs to design a vaccine that is capable of eliciting cross‐protective immune responses (as next‐gen broad‐spectrum vaccines).

The AI‐ML advances have substantially improved antigen assessment by enabling rapid prediction of protein structure and computational identification of the epitope. AlphaFold2 developed by DeepMind has demonstrated high‐precision atomic‐level accuracy in predicting 3D protein structures, thereby facilitating recognizable proof of preserved locales as appropriate immunization targets. Complementing this, the deep‐learning EpitopePredikt analyses and accurately predicts B‐cell and T‐cell epitopes for improved immunological targeting. Another ML‐based tool Vaxign‐ML adopts a switch vaccinology approach to formulate defensive antigens by deciphering the genome of a target pathogen that subsequently returned the identifiable proof of useful vaccine candidates [29]. The dynamic existence of SARS‐CoV‐2 characterized and facilitated by frequent mutations necessitates adaptable, innovative, and inventive vaccine strategies. AI‐driven platforms facilitate rapid evolution of booster vaccines to match with evolving spike protein epitopes. Vaxformer, a transformation‐based model, generates antigenicity‐controlled spike protein sequences, enabling vaccine designing to elicit high immune responses against emerging variants [30]. Emerging AI models aim to evaluate plausible evolutionary trajectories of viral proteins and might assist in identifying antigen alterations that warrant further investigation. Due to the complex interplay of the evolutionary, immunological, and epidemiological factors, accurate evolutionary prediction of viruses in the future still remains a challenge.

Studies showed that developing a vaccine by integrating “conserved domain” non‐spike antigen gave broader protection and enhanced T‐cell response [31]. Biopharmaceutical companies like Baseimmune Ltd., UK, leverage AI to predict mutations in pathogens and develop vaccine that would remain effective over an extended period and reduce frequent booster dosing [32]. AI‐ML in vaccine development also extends to the discovery of adjuvants and optimized vaccine formulation. In silico platforms utilize AI to predict novel adjuvant candidates and assess antigen‐specific compatibility, expediting safe and effective formulated vaccine development [33]. AI models also assist in optimizing lipid nanoparticles (LNPs and iLNPs) formulation as mRNA vaccine delivery vehicles, enhancing its stability and immunogenicity. As a cutting‐edge advanced technique, AI‐ML application in vaccine development faces numerous challenges. The accuracy of AI predictions is contingent upon the quality and diversity of the training datasets, necessitating comprehensive and representative data collection. Additionally, integrating AI tools into the existing vaccine development framework needs broad interdisciplinary collaboration and regulatory considerations for safety and efficacy. Emerging and reemerging variants call for robust, agile, and flexible vaccine development platforms. DL and ML have transformative potential to accelerate vaccine design, optimize antigen and adjuvant selection and anticipate viral mutations. Global healthcare community and agencies shall have enhanced preparedness and response to the current and future pandemic threats to ultimately mitigate the adverse impacts on public healthcare infrastructure.

7. Public Health and Surveillance Strategies

Recent emerging and reemerging SARS‐CoV‐2 variants LF.7 and NB.1.8.1 vindicate ongoing research and development involving cutting‐edge technology and techniques in vaccinology, community health, “one health” and total health solutions [34]. Flexible mRNA vaccine platform and AI‐ML‐based in silico approaches are few such state‐of‐art interventions. Characterized by transformations, it would impact transmissibility and infectivity. Multifaceted and coordinated AI strategies, periodic wastewater analyses and real‐time genomic surveillance could help to successfully counter the variants [35]. ML technique allows analyze large datasets to predict a mutation, and facilitates planned antigen‐antibody designing to tackle emerging variants [36]. Such capabilities shall provide upgraded details of the antibodies, thus guaranteeing better scope to time‐to‐time counter circulating strains. AI could help to optimize booster rollout protocol by differentiating the high‐risk populations and anticipating potential hotspots, leading to a focused immunization strategy [37]. Microbial analysis of wastewaters to identify emerging variants (like LF.7 and NB.1.8.1) is well acknowledged [38]. Monitoring and surveillance of wastewaters are important for early signs of circulating strains in a community. Community healthcare authorities could identify the nearness of the viral RNA by analyzing sewage microbiology. This cost‐effective strategy is especially useful to observe asymptomatic spread to follow the disease dynamics.

Real‐time genomic surveillance is the basis to understand advances and spread of a variant. Analysts could identify the genomic alterations (like high transmissibility, vaccine resistance, etc.) that could influence infection patterns, by sequencing the viral genomes at clinic level. Such data are significant in favored resource mobilization, countermeasure planning and execution and altering the conventional treatment measures. Combining genomic data with epidemiological data shall upgrade the ability to follow transmission chains and effectively contain a brewing outbreak [39]. Investing in comprehensive surveillance systems to maintain and sustain control over the highly‐mutating COVID‐19 virus that causes pandemic‐like health threats wave after wave. Exemplified by LF.7 and NB.1.8.1 subvariants, the ongoing evolved SARS‐CoV‐2 highlights the necessity of adaptive and integrated surveillance strategy. Robust surveillance framework capable to respond swiftly to dynamic challenges by novel variants may be designed by AI‐driven analytics, wastewater surveillance and genome monitoring combinations. Integrated approach could help predict an outbreak, make informed vaccination drive decisions and implement targeted healthcare interventions. Continued investment and collaborations in these shall enhance our ability to manage current and future infectious disease challenges and threats.

8. Conclusion

As conventional surveillance has its limitations, the ongoing onslaught of COVID‐19 subvariants highlights the necessity of employing state‐of‐art technologies. AI‐ML platforms like Nextstrain, Pangolin‐ML, federated frameworks, and deep learning models made it possible to rapidly identify mutations, assign lineage, and real‐time global genomics monitoring. Combining genome, epidemiology, and structural data, they improve transmission prediction and high‐risk variant prioritization. AI‐ML expedites vaccine development by designing optimal immunogen, optimize sequences, and predict epitope, as demonstrated in mRNA vaccines. Thus, it was essential to integrate transparent, morally sound, and trustworthy AI‐ML models for improved surveillance, quick response, and vaccine development against the altered pandemic scenario. The observations and perspectives this article has put forth shall be a step forward in addressing even other emerging infectious disease outbreaks with pandemic potential.

Author Contributions

Aurobinda Rout: data curation, writing – original draft, investigation. Kumarjit Das: data curation, writing – original draft, investigation. Ashish K. Sarangi: data curation, writing – original draft. Snehasish Mishra: conceptualization, validation, supervision, writing – review and editing. Chandana Mohanty: data curation, writing – original draft. Puneet K. Singh: data curation, writing – original draft, validation. Swikrutee Rout: data curation, writing – original draft. Ranjan K. Mohapatra: project administration, writing – original draft, methodology. Lawrence Sena Tuglo: validation, writing – review and editing.

Funding

The authors have nothing to report.

Ethics Statement

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Transparency Statement

The corresponding authors (L.S. Tuglo and S. Mishra) affirm that the article is an honest, accurate, and transparent account of the reported study; that no important aspects of the study are omitted; and that any discrepancies from the study as planned (and, if relevant, registered) are explained.

Rout A., Das K., Sarangi A. K., et al., “Emerging Novel SARS‐CoV‐2 Subvariants and the Advantages of AI‐ML in Deciphering the Mutation Trends, Genomic Surveillance and Vaccine Development,” Health Science Reports 9 (2026): e72792, 10.1002/hsr2.72792.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.