About

Joseph S.B. Mitchell is a SUNY Distinguished Professor in the Department of Applied Mathematics and Statistics at the State University of New York at Stony Brook. His research interests encompass computational geometry, algorithms and data structures, optimization, operations research, graphics and visualization, as well as computer-aided (geometric) design and manufacturing. He is associated with the AI Innovation Institute at Stony Brook, contributing to advancements in these fields through his academic and research activities.

Research topics

• Computer Science
• Artificial Intelligence
• Discrete mathematics
• Mathematics
• Mathematical analysis
• Computer vision
• Algorithm
• Programming language
• Combinatorics
• Geometry
• Distributed computing

Selected publications

• Influence of Lipid Composition on Nonspecific Interactions of Serotonin with Model Membranes

  ACS Pharmacology & Translational Science · 2026-01-29

  articleOpen access

  Serotonin is a monoamine neurotransmitter, which plays an important role in the development and functioning of the central nervous system. Recent biophysical studies reveal that nonspecific interactions between serotonin and lipid membranes significantly alter lipid bilayer properties, impacting synaptic function and plasticity. To better understand these critical interactions and their broader implications for neural function and pharmacology, we investigated the interactions of serotonin (at concentrations ranging from 1 to 40 mM) with model membranes prepared as droplet interface bilayers, liposomes, and supported bilayers. These membrane systems comprised single, binary, and ternary lipid mixtures, including pure DOPC, DOPC/DOPS (10:1 mol ratio), and DOPC/Sphingomyelin/Cholesterol (1:1:0.2 mol ratio). Our analysis employing various experimental techniques shows that the interaction of serotonin with lipid membranes of diverse compositions has overall nonspecific effects in (1) influencing the barrier properties of the lipid membrane, as demonstrated by increased water permeability compared to the control; (2) modifying the phase transition behavior, evidenced by decrease in the main phase transition temperature and reduction of the transition enthalpy; (3) perturbing the conformational ordering of lipid membranes, as indicated by the increase in specific Raman intensity ratio; and (4) reducing bilayer tension with increasing serotonin concentrations. Overall, membrane modifications increase with rising serotonin concentrations, plateauing at higher levels. Sensitivity to serotonin varies by lipid composition in the order: DOPC/DOPS ≈ DOPC/Sphingomyelin/Cholesterol > DOPC. Our experimental findings reveal that serotonin significantly alters membrane properties, particularly affecting neuronal membrane composition and lipid rafts, which are critical for membrane protein organization and signaling. These findings suggest that serotonergic drugs and pathological fluctuations in serotonin may influence signaling not only through classical receptor-mediated pathways, but also by altering the lipid-protein landscape of the membrane, with potential implications for drug efficacy, off-target effects, and the development of therapies that target membrane composition in serotonin-related disorders.

  PublisherDOI

• Membrane-Modifying Effects of Perfluoroalkyl Substances in Model Bacterial Membranes

  ACS Omega · 2025-08-26 · 5 citations

  articleOpen access

  Understanding the interactions of per- and polyfluoroalkyl substances (PFAS) with bacterial membranes is essential for evaluating their ecological and health impacts. To mimic the diverse environments found in bacterial membranes, we constructed model membranes as bilayers, liposomes, and supported bilayers using binary lipid mixtures of 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) or 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE), with 1,2-dioleoyl-sn-glycero-3-phospho-(1′-rac-glycerol) (sodium salt) (DOPG)─all sharing the same acyl chains but differing in headgroup types and charges. Our findings demonstrate that salts of perfluorooctanoic acid (PFOA) and perfluorobutanesulfonic acid (PFBS) induce concentration- and lipid-dependent disordering effect in membranes composed of DOPC/DOPG (3:1 mol ratio) and DOPE/DOPG (3:1 mol ratio). Water permeability measurements reveal that membranes with greater hydrogen bonding capacity and curvature stress─such as those containing DOPE─exhibit more pronounced increases in permeability upon PFAS exposure, indicating heightened susceptibility to disruption by these contaminants. Differential scanning calorimetry (DSC) shows that DOPE-DOPG mixtures display a more significant decrease in phase transition temperature (Tm) and enthalpy compared to DOPC-DOPG membranes. Moreover, Raman and attenuated total reflectance infrared (ATR-IR) spectroscopies reveal a greater increase in lipid acyl chain disorder in DOPE-DOPG mixtures upon PFAS exposure. Collectively, these findings indicate that PFAS salts not only increase membrane permeability but also destabilize lipid packing and phase organization, with the most pronounced disordering effects observed in membranes containing DOPE. Taken together, our results highlight the complex interplay of electrostatic, van der Waals, and hydrogen bonding interactions that govern the effects of PFAS salts on bacterial membrane properties, as revealed by their differential impacts on permeability and lipid organization.

  PublisherDOI

• Provable Methods for Searching with an Imperfect Sensor

  2025-05-19 · 1 citations

  articleSenior author

  Assume that a target is known to be present at an unknown point among a finite set of locations in the plane. We search for it using a mobile robot that has imperfect sensing capabilities. It takes time for the robot to move between locations and search a location; we have a total time budget within which to conduct the search. We study the problem of computing a search path/strategy for the robot that maximizes the probability of detection of the target. Considering non-uniform travel times between points (e.g., based on the distance between them) is crucial for search and rescue applications; such problems have been investigated to a limited extent due to their inherent complexity. In this paper, we describe fast algorithms with performance guarantees for this search problem and some variants, complement them with complexity results, and perform experiments to characterize their performance.

  PublisherDOI

• Voluntary mobility clustering for epidemic control

  2025-11-03

  articleOpen access

  In case of a future pandemic, the mobility dynamics of a city can be controlled by intervening in the mobility patterns of people. Instead of hard quarantine policies, incentives can be designed that are compatible with people's preferences. At first, we distinguish mobility from the different types of locations for which distance matters. We match these types of locations in a way that maximizes the natural preference of people to visit the locations. We investigate different approaches for matching locations, such as retail and educational services, while considering people's preferences. We show that satisfying the preferences of the entire city is a computationally hard problem. Approximation algorithms are proposed in which the penalty for preference violation is bounded. We propose a fast approximation algorithm that focuses on the penalty value of locations, and we propose a more computationally heavy approximation that focuses on user penalty with a specific scheme of user allocation to locations. Additionally, we investigated higher-order matching of locations and the complexity of urban partitioning. We tested our approach in Euclidean space and network space. Finally, we show that applying such mobility restrictions can reduce the transmission rate, and we extract cells whose people can be incentivized to fulfill their needs based on the proposed algorithms, slowing down a future pandemic and preventing potential superspreading events.

  PublisherOA PDFDOI

• Vantage Point Selection Algorithms for Bottleneck Capacity Estimation

  ArXiv.org · 2025-01-01

  preprintOpen access

  Motivated by the problem of estimating bottleneck capacities on the Internet, we formulate and study the problem of vantage point selection. We are given a graph $G=(V, E)$ whose edges $E$ have unknown capacity values that are to be discovered. Probes from a vantage point, i.e, a vertex $v \in V$, along shortest paths from $v$ to all other vertices, reveal bottleneck edge capacities along each path. Our goal is to select $k$ vantage points from $V$ that reveal the maximum number of bottleneck edge capacities. We consider both a non-adaptive setting where all $k$ vantage points are selected before any bottleneck capacity is revealed, and an adaptive setting where each vantage point selection instantly reveals bottleneck capacities along all shortest paths starting from that point. In the non-adaptive setting, by considering a relaxed model where edge capacities are drawn from a random permutation (which still leaves the problem of maximizing the expected number of revealed edges NP-hard), we are able to give a $1-1/e$ approximate algorithm. In the adaptive setting we work with the least permissive model where edge capacities are arbitrarily fixed but unknown. We compare with the best solution for the particular input instance (i.e. by enumerating all choices of $k$ tuples), and provide both lower bounds on instance optimal approximation algorithms and upper bounds for trees and planar graphs.

  PublisherOA PDFDOI

• P655: Improved representation of functional data in ClinVar

  Genetics in Medicine Open · 2025-01-01

  articleOpen access

  Methods: We recruited patients with suspected genetic disorders and conducted comprehensive evaluations, including medical histories, physical exams, and relevant diagnostic tests.Initial molecular and cytogenetic testing performed at clinical laboratories either yielded negative results or identified variants of uncertain significance.As a result, analysis of genome sequencing and RNA-seq data, including splicing evaluation, were performed at the University of Utah and ARUP Laboratories to aid in these patients' diagnostic journeys.Results: The study analyzed 20 patients (9 males, 11 females) with various clinical presentations.RNA-seq analysis provided functional data that helped confirm diagnoses or identify likely diagnoses in eight cases (3 males and 5 females).Specifically, RNA-seq played a pivotal role in upgrading variant pathogenicity in seven cases, including three with intronic splicing variants identified by genome sequencing and confirmed as pathogenic through RNA-seq.Case 1: A 16-month-old boy with suspected Diamond-Blackfan Anemia (DBA) presented with severe anemia, low reticulocytes, and elevated erythrocyte adenosine deaminase (eADA) levels, requiring frequent transfusions.Despite initial negative genetic testing, genome sequencing revealed a novel maternally inherited splicing variant in RPS7 (c.-19G>C), which was predicted to impair splicing.RNA-seq confirmed significant intron retention in RPS7 transcripts in both the proband and his mother, indicating a strong allelic association.The variant was classified as pathogenic and consistent with DBA8.The patient underwent a successful hematopoietic stem cell transplant (HSCT) from an HLA-matched sibling who did not carry the variant RPS7, achieving transfusion independence for over 8 months with 100% donor chimerism and currently normal hematologic indices.Case 2: A 20-month-old girl presented with severe anemia, red cell aplasia, and congenital heart defects.Initial genetic testing was negative, but genome sequencing identified a de novo intronic RPS19 variant (c.172+350C>T), predicted to create a novel splice donor site.RNA-seq confirmed the presence of an 81-bp novel exon downstream of RPS19 exon 3 in about half of the proband's transcripts, absent in controls.This variant was classified as pathogenic based on its de novo occurrence, phenotype match, absence from population databases, and functional evidence.This is consistent with a diagnosis of DBA1.The patient is preparing for HSCT with a 12/12 HLA-matched sibling donor, who does not carry the RPS19 variant.Case 3: A 10-year-old girl presented with short stature, metaphyseal irregularities around the knee, vertebral abnormalities, and rib widening, suggestive of a spondyloepiphyseal dysplasia.Initial genetic testing, including a 54-gene skeletal dysplasia panel and exome sequencing, was negative.Genome sequencing identified compound heterozygous splicing variants in EFL1: a maternally inherited likely pathogenic variant (c.932+2T>G) and a paternally inherited variant of uncertain significance (c.245-12A>G).RNA-seq confirmed splicing defects, including exon skipping for c.932+2T>G and intron retention for c.245-12A>G, supporting the pathogenicity of both variants and confirming a diagnosis of Schwachman-Diamond syndrome 2. Conclusion: This study underscores the value of RNA-seq in unraveling complex diagnostic cases, particularly for intronic variants of uncertain significance.By integrating RNA-seq with genome sequencing, we have enhanced our ability to identify elusive genetic causes of disease.This combined approach proves crucial for resolving diagnostic challenges, establishing RNA-seq and DNA sequencing as indispensable tools in clinical genetics for rare and undiagnosed diseases.

  PublisherOA PDFDOI

• On some geometric optimization problems with segments

  Theoretical Computer Science · 2025-06-04

  article1st author
  PublisherDOI

• Multi-robot Searching with Limited Sensing Range for Static and Mobile Intruders

  2025-08-18

  article

  We consider the problem of searching for an intruder in a geometric domain by utilizing multiple search robots. The domain is a simply connected orthogonal polygon with edges parallel to the coordinate axes. Each robot has a limited sensing capability. We study the problem for both static and mobile intruders. It turns out that the problem of finding an intruder is NP-hard, even for a stationary intruder. Given this intractability, we turn our attention towards developing efficient and robust algorithms, namely methods based on space-filling curves, random search, and cooperative random search. Moreover, for each proposed algorithm, we evaluate the trade-off between the number of search robots and the time required for the robots to complete the search process.

  PublisherDOI

• Guarding Offices with Maximum Dispersion

  ArXiv.org · 2025-06-26

  preprintOpen access

  We investigate the Dispersive Art Gallery Problem with vertex guards and rectangular visibility ($r$-visibility) for a class of orthogonal polygons that reflect the properties of real-world floor plans: these office-like polygons consist of rectangular rooms and corridors. In the dispersive variant of the Art Gallery Problem, the objective is not to minimize the number of guards but to maximize the minimum geodesic $L_1$-distance between any two guards, called the dispersion distance. Our main contributions are as follows. We prove that determining whether a vertex guard set can achieve a dispersion distance of $4$ in office-like polygons is NP-complete, where vertices of the polygon are restricted to integer coordinates. Additionally, we present a simple worst-case optimal algorithm that guarantees a dispersion distance of $3$ in polynomial time. Our complexity result extends to polyominoes, resolving an open question posed by Rieck and Scheffer (CGTA 2024). When vertex coordinates are allowed to be rational, we establish analogous results, proving that achieving a dispersion distance of $2+\varepsilon$ is NP-hard for any $\varepsilon > 0$, while the classic Art Gallery Problem remains solvable in polynomial time for this class of polygons. Furthermore, we give a straightforward polynomial-time algorithm that computes worst-case optimal solutions with a dispersion distance of $2$. On the other hand, for the more restricted class of hole-free independent office-like polygons, we propose a dynamic programming approach that computes optimal solutions. Moreover, we demonstrate that the problem is practically tractable for arbitrary orthogonal polygons. To this end, we compare solvers based on SAT, CP, and MIP formulations. Notably, SAT solvers efficiently compute optimal solutions for randomly generated instances with up to $1600$ vertices in under $15$s.

  PublisherOA PDFDOI

• The value of international pharmacovigilance databases to help contain antibiotic resistance from a 'one health' perspective

  2025-11-03

  articleOpen access1st authorCorresponding

  <p dir="ltr">Background</p><p dir="ltr">Antibiotic resistance (ABR) is a global threat to public health that requires action from a 'One health' perspective. 'One health' highlights the interlinked nature of human, animal and environmental health. ABR can occur within humans, animals and the environment and can transmit between these different areas, thus the need to incorporate 'One health' into efforts to contain ABR. However, the relative importance of different drivers within human, animal and environmental health is not known. A greater understanding of this may allow more targeted interventions across 'One health'. Furthermore, limitations within traditional, laboratory-based surveillance of ABR have led to calls to consider other data sources. One of these hypothesised sources is data from pharmacovigilance databases. Pharmacovigilance databases collect reports of suspected adverse events related to the use of medicines. Pharmacovigilance benefits from a well- established network, with a particularly large reach for human health. This thesis aimed to assess the relative associations of different aspects of 'One health' in the ABR of children from rural India. Furthermore, this thesis aimed to investigate the utility of pharmacovigilance databases as a supplementary tool to that of traditional ABR surveillance across 'One health'.</p><p dir="ltr">Methods</p><p dir="ltr">Paper I was a cohort study that used Escherichia coli isolated from the stools of 125 children (aged between one and three years old) and animals, as well as water sources within the same community. Samples were collected simultaneously across sources at seven time points over a two year period and susceptibility testing for 16 antibiotics performed. Newey-West regression models were used to study temporal associations between resistance levels in humans and that of other sources. SourceR was used for attribution modelling to estimate the proportion of ABR cases in humans which could be attributed to each of the other sources.</p><p dir="ltr">Paper II analysed potential reports of ABR in VigiBase, a global pharmacovigilance database for humans. The reports were split into 'Probable' and 'Possible' cases of ABR based on the reported adverse events in the report. All 'Probable' reports and a random selection of 599 'Possible' reports were reviewed on a case-by- case basis. The reports were reviewed for their likelihood to be a case of ABR, as well as the reporting of antimicrobial susceptibility testing, previous antibiotic use, if the cases were linked to published scientific literature and the demographics of the reported patients. Chi-square tests and logistic regression models were used to test for differences between the 'Probable' and 'Possible' reports.</p><p dir="ltr">Paper III studied the reporting of potential ABR cases in international pharmacovigilance databases across 'One health'. VigiBase and EudraVigilance Veterinary, a European-based animal health pharmacovigilance database, were used to identify cases for human, animal and environmental health. No specialist environmental health database was identified. The reports were analysed over time, by continent of the reporting country and by the type of antibiotic reported using the Anatomical Therapeutic Chemical (ATC) classification.</p><p dir="ltr">In paper IV we compared the reports of potential ABR cases in VigiBase and EudraVigilance Veterinary against the percentage of resistant tests reported to traditional ABR surveillance databases. For humans, the Global Antimicrobial Resistance and Use Surveillance System (GLASS) was used and for animals it was the European Food Safety Authority (EFSA) and European Centre for Disease Prevention and Control (ECDC) joint report on ABR. The most recent data from these ABR databases was from 2022, and the data from the pharmacovigilance databases was limited to all reports up to the end of 2022. Logistic regression models were used to compare how ABR was reported for different ATC subgroups between the different databases.</p><p dir="ltr">Results</p><p dir="ltr">Eighteen of the 69 Newey-West regression models estimating the temporal association between human ABR and that of the other sources were statistically significant. Most of these, eight, were with household drinking water. There were five, three and two statistically significant temporal relationships for wastewater, animals and source of drinking water, respectively. SourceR attribution modelling estimated that the highest proportion of attributable cases of human ABR were from household drinking water, animals and wastewater. The proportion of attributable cases of human ABR was much lower for the source of drinking water.</p><p dir="ltr">There were more 'Possible' reports of potential ABR in VigiBase (n=20,815) compared to 'Probable' (n=3,497). For both groups, most of the reports were from high-income countries. The 'Probable' reports were more likely to be cases of ABR and report antimicrobial susceptibility testing. The logistic regression models showed that the 'Probable' reports were more likely to be cases of ABR after accounting for potential confounders. However, the 'Possible' reports were more likely to be reported from non-healthcare professionals and contained valuable information regarding patients who experienced an unexpected lack of effect of antibiotic treatment.</p><p dir="ltr">In terms of total reports, there were more reports for humans (n=29,667) compared to animals (n=5,217) in international pharmacovigilance databases. There were no reports reporting both environment and ABR terms together in either VigiBase or EudraVigilance Veterinary. There were 52 reports of environment-related terms with an antibiotic. There was an increase in reports received annually over time for humans and animals, but this was less clear for environment-related reporting. There was a consistent difference, for at least one antibiotic ATC subgroup, between the antibiotics reported to pharmacovigilance databases for potential cases of ABR and the percentage of resistant tests for these antibiotic groups in specialised databases.</p><p dir="ltr">Conclusion</p><p dir="ltr">'One health' should be considered when it comes to ABR surveillance and interventions. In the rural Indian community we studied, the closest over time association was with household drinking water. Pharmacovigilance databases can be used to identify potential cases of ABR from around the globe, but the number of reports of and the scope of the database was greatest for humans followed by animals. There was an absence of a specialised database for environment-related reporting. For humans, the search of VigiBase was able to identify cases of ABR with reasonable accuracy, and this was more so for a more specific search. When compared to traditional ABR databases, there was a consistent difference between the antibiotics in the human and animal-based databases.</p><h3>List of scientific papers</h3><p dir="ltr">I. <b>Mitchell J,</b> Purohit M, Jewell CP, Read JM, Marrone G, Diwan V, Stålsby Lundborg C. Trends, relationships and case attribution of antibiotic resistance between children and environmental sources in rural India. Sci Rep. 2021 Nov 19;11(1):22599. <a href="https://doi.org/10.1038/s41598-021-01174-w" rel="noreferrer" target="_blank">https://doi.org/10.1038/s41598-021-01174-w</a></p><p dir="ltr">II. <b>Mitchell J,</b> Westerberg C, Purohit M, Lundquist P, Stålsby Lundborg C. The Enhancing Role of Pharmacovigilance to Conventional Antibiotic Resistance Surveillance: Cross-sectional Identification and Analysis of Reports of Antibiotic Resistance in VigiBase. Int J Infect Dis. 2025 June 4;107947. <a href="https://doi.org/10.1016/j.ijid.2025.107947" rel="noreferrer" target="_blank">https://doi.org/10.1016/j.ijid.2025.107947</a></p><p dir="ltr">III. <b>Mitchell J,</b> Lundquist P, Westerberg C, Purohit M, Stålsby Lundborg C. A 'One Health' Analysis of Reports of Potential Antibiotic Resistance Cases in International Pharmacovigilance Databases. [Submitted]</p><p dir="ltr">IV. <b>Mitchell J,</b> Purohit M, Lundquist P, Westerberg C, Stålsby Lundborg C. A Comparison of Antibiotic Resistance Reports in Pharmacovigilance Databases to Conventional Surveillance Across 'One Health'. [Submitted]</p>

  PublisherDOI

Recent grants

• Algorithmic Studies in Applied Geometry

  NSF · $200k · 2007–2010

• AF: Small: Approximation Algorithms for Geometric Network Optimization

  NSF · $451k · 2015–2019

• International Symposium on Computational Geometry (SOCG) 2015, Eindhoven, The Netherlands, June 22-25, 2015

  NSF · $15k · 2015–2016

• AF: Small: Approximation Algorithms for Geometric Optimization

  NSF · $483k · 2010–2014

• AF:Small:Geometric Optimization Problems for Routing, Searching, and Coverage in the Face of Uncertainty

  NSF · $450k · 2020–2024

Frequent coauthors

• Esther M. Arkin

  Hangzhou Dianzi University

  219 shared

• Valentin Polishchuk

  85 shared

• Sándor P. Fekete

  Technische Universität Braunschweig

  83 shared

• Steven Skiena

  53 shared

• Michael A. Bender

  Stony Brook University

  43 shared

• Alon Efrat

  Alexandru Ioan Cuza University

  35 shared

• Martin Held

  33 shared

• Erik D. Demaine

  33 shared