TacticAI: an AI assistant for football tactics.

Identifying key patterns of tactics implemented by rival teams, and developing effective responses, lies at the heart of modern football. However, doing so algorithmically remains an open research challenge. To address this unmet need, we propose TacticAI, an AI football tactics assistant developed and evaluated in close collaboration with domain experts from Liverpool FC. We focus on analysing corner kicks, as they offer coaches the most direct opportunities for interventions and improvements. TacticAI incorporates both a predictive and a generative component, allowing the coaches to effectively sample and explore alternative player setups for each corner kick routine and to select those with the highest predicted likelihood of success. We validate TacticAI on a number of relevant benchmark tasks: predicting receivers and shot attempts and recommending player position adjustments. The utility of TacticAI is validated by a qualitative study conducted with football domain experts at Liverpool FC. We show that TacticAI's model suggestions are not only indistinguishable from real tactics, but also favoured over existing tactics 90% of the time, and that TacticAI offers an effective corner kick retrieval system. TacticAI achieves these results despite the limited availability of gold-standard data, achieving data efficiency through geometric deep learning.

Mastering the game of Stratego with model-free multiagent reinforcement learning.

We introduce DeepNash, an autonomous agent that plays the imperfect information game Stratego at a human expert level. Stratego is one of the few iconic board games that artificial intelligence (AI) has not yet mastered. It is a game characterized by a twin challenge: It requires long-term strategic thinking as in chess, but it also requires dealing with imperfect information as in poker. The technique underpinning DeepNash uses a game-theoretic, model-free deep reinforcement learning method, without search, that learns to master Stratego through self-play from scratch. DeepNash beat existing state-of-the-art AI methods in Stratego and achieved a year-to-date (2022) and all-time top-three ranking on the Gravon games platform, competing with human expert players.

Multiagent off-screen behavior prediction in football.

In multiagent worlds, several decision-making individuals interact while adhering to the dynamics constraints imposed by the environment. These interactions, combined with the potential stochasticity of the agents' dynamic behaviors, make such systems complex and interesting to study from a decision-making perspective. Significant research has been conducted on learning models for forward-direction estimation of agent behaviors, for example, pedestrian predictions used for collision-avoidance in self-driving cars. In many settings, only sporadic observations of agents may be available in a given trajectory sequence. In football, subsets of players may come in and out of view of broadcast video footage, while unobserved players continue to interact off-screen. In this paper, we study the problem of multiagent time-series imputation in the context of human football play, where available past and future observations of subsets of agents are used to estimate missing observations for other agents. Our approach, called the Graph Imputer, uses past and future information in combination with graph networks and variational autoencoders to enable learning of a distribution of imputed trajectories. We demonstrate our approach on multiagent settings involving players that are partially-observable, using the Graph Imputer to predict the behaviors of off-screen players. To quantitatively evaluate the approach, we conduct experiments on football matches with ground truth trajectory data, using a camera module to simulate the off-screen player state estimation setting. We subsequently use our approach for downstream football analytics under partial observability using the well-established framework of pitch control, which traditionally relies on fully observed data. We illustrate that our method outperforms several state-of-the-art approaches, including those hand-crafted for football, across all considered metrics.

Peptide Enhanced Bone Graft Substitute Presents Improved Short-Term Increase in Bone Volume and Construct Stiffness Compared to Iliac Crest Autologous Bone in an Ovine Lumbar Interbody Fusion Model.

STUDY DESIGN: Preclinical ovine model. OBJECTIVE: To assess the in vivo efficacy and safety of the P-15 L bone graft substitute and compare its performance to autologous iliac crest bone graft (ICBG) for lumbar interbody fusion indications. METHODS: Thirty skeletally mature sheep underwent lumbar interbody fusion surgery. Half of the sheep received autologous ICBG and the other half the peptide enhanced bone graft substitute (P-15 L). Following termination at 1, 3, and 6 months after surgery, the operated segments were analyzed using micro computed tomography (µCT), histology, and destructive mechanical testing. Additional systemic health monitoring was performed for the P-15 L group. RESULTS: One month after surgery, there was only minor evidence of bone remodeling and residual graft material could be clearly observed within the cage. There was active bone remodeling between 1 and 3 months after surgery. At 3 months after surgery significantly denser and stiffer bone was found in the P-15 L group, whereas at 6 months, P-15 L and ICBG gave similar fusion results. The P-15 L bone graft substitute did not have any adverse effects on systemic health. CONCLUSIONS: The drug device combination P-15 L was demonstrated to be effective and save for lumbar interbody fusion as evidenced by this ovine model. Compared to autologous ICBG, P-15 L seems to expedite bone formation and remodeling but in the longer-term fusion results were similar.

Navigating the landscape of multiplayer games.

Multiplayer games have long been used as testbeds in artificial intelligence research, aptly referred to as the Drosophila of artificial intelligence. Traditionally, researchers have focused on using well-known games to build strong agents. This progress, however, can be better informed by characterizing games and their topological landscape. Tackling this latter question can facilitate understanding of agents and help determine what game an agent should target next as part of its training. Here, we show how network measures applied to response graphs of large-scale games enable the creation of a landscape of games, quantifying relationships between games of varying sizes and characteristics. We illustrate our findings in domains ranging from canonical games to complex empirical games capturing the performance of trained agents pitted against one another. Our results culminate in a demonstration leveraging this information to generate new and interesting games, including mixtures of empirical games synthesized from real world games.

Autologous fat processing via the Revolve system: quality and quantity of fat retention evaluated in an animal model.

BACKGROUND: Currently, fat graft viability and retention cannot be reliably predicted. The reasons for this variability are not fully understood, although fat processing has been implicated. OBJECTIVES: The authors compare the in vitro quantity and in vivo fat retention from lipoaspirate processed by the Revolve system (LifeCell, Bridgewater, New Jersey) compared with centrifugation and decantation. METHODS: Ten patients were enrolled in this prospective study. Lipoaspirate from each patient was processed by each of 3 methods: decantation, centrifugation, and the Revolve system. Biochemical characteristics and free oil, adipose, and aqueous phases of the processed fats were determined. Fat grafts were implanted in nude mice; volume retention and quality of the fat grafts were evaluated after 28 days. Viability of retained fat was demonstrated by intact adipocytes and neovascularization on histology. RESULTS: Of the 10 patients, 9 were women and 1 was a man. Mean patient age was 40.7 ± 8.9 years (range, 30-55 years). Fat tissue obtained from all methods had good physiological properties with neutral pH and isotonic salt concentrations. The Revolve system yielded significantly less blood cell debris, a higher percentage of adipose tissue, and a lower percentage of free oil compared with the other 2 methods. Fat tissue retention from Revolve samples was significantly higher (73.2%) than that from decanted samples (37.5%) and similar to that from centrifuged samples (67.7%). CONCLUSIONS: The Revolve system produced physiologically compatible, preinjection fat with reduced contaminants and free oil in conjunction with high fat content. In an animal model, volume retention of Revolve-processed fat grafts was significantly greater than decanted samples. The Revolve system presents a fat-processing option that was less time-consuming, easier to use, and more efficient in this study than standard centrifugation or decantation.

Implantation of a porcine acellular dermal graft in a primate model of rotator cuff repair.

BACKGROUND: Non-cross-linked xenogeneic extracellular matrix graft materials have typically elicited a hypersensitivity reaction when implanted into humans or other primates. The purpose of this study was to examine the histologic and immune response to a non-cross-linked porcine-derived dermal extracellular matrix graft processed to remove the α-gal epitope. MATERIALS AND METHODS: Eight African green monkeys were implanted with porcine acellular dermal matrix (Conexa Reconstructive Tissue Matrix; Tornier Inc, Edina, MN, USA) to repair and augment a partial excision defect of the supraspinatus tendon of the rotator cuff. Four animals each were sacrificed at 3 months and 6 months, and histologic samples were compared with tissues harvested from unoperated shoulders. RESULTS: Gross examination of grafted Conexa showed the appearance of integration proximally with tendon and distally with bone in each operated rotator cuff complex. Histologically, Conexa appeared to have remodeled to tendon-like architecture, with homogeneous distribution of fibroblast cells and parallel alignment of collagen fibers, with the direction of force evident by 3 months after implantation. Abundant vasculature observed at 3 months, which diminished to native tendon levels by 6 months, also indicated this to be a period of significant remodeling with an absence of significant inflammation, as evidenced by immunochemical methods and serum analysis. CONCLUSION: Conexa porcine acellular dermal matrix allows for incorporation of host tendon tissue without a hypersensitivity reaction in a primate model and should be a safe material for augmentation of human rotator cuff repair.

The use of acellular dermal matrix to prevent capsule formation around implants in a primate model.

BACKGROUND: Implant-based breast reconstruction is a popular option after mastectomy, but capsular contracture may detract from long-term outcomes. The authors have observed that breast implants covered with acellular dermal matrix (AlloDerm) are less likely to develop a capsule in the area where the implant is in direct contact with the acellular matrix. The authors tested this observation experimentally by comparing capsular formation around implants in the presence and absence of AlloDerm in primates. METHODS: Eight smooth-surfaced tissue expanders were implanted into eight African green monkeys. In four experimental animals, a sheet of AlloDerm was draped over the tissue expander so as to cover the implant. Four control animals underwent placement of a tissue expander only. Animals were killed after 10 weeks and specimens underwent histologic and immunohistochemical analysis. RESULTS: Hematoxylin and eosin staining of control specimens revealed the presence of a distinct layer of wavy, parallel arrays of collagen fibers consistent with capsule formation. Immunostaining identified abundant myofibroblasts, a profibrotic cell found in breast capsules. In the AlloDerm-covered specimens, no capsule layer was visible, and specimens stained weakly for myofibroblasts. The difference in myofibroblast staining intensity was statistically significant. CONCLUSIONS: The use of AlloDerm to partially enclose implants effectively prevented formation of a capsule in areas where AlloDerm contacted the implant at 10 weeks. Long-term studies will be required to determine whether this is a durable result that can be reproduced in humans.

Inactivation of bacterial spores and viruses in biological material using supercritical carbon dioxide with sterilant.

The purpose of this study was to validate supercritical carbon dioxide (SC-CO(2)) as a terminal sterilization method for biological materials, specifically acellular dermal matrix. In this study, bacterial spores, Bacillus atrophaeus, were inoculated onto porcine acellular dermal matrix to serve as a "worst case" challenge device. The inactivation of the spores by SC-CO(2) with peracetic acid (PAA) sterilant was analyzed as a function of exposure times ranging from 1 to 30 min. A linear inactivation profile for the Bacillus atrophaeus spores was observed, and a SC-CO(2) exposure time of 27 min was determined to achieve a sterility assurance level of 10(-6). The inactivation of viruses was also studied using Encephalomyocarditis (EMC) viruses. After 15 min of exposure to SC-CO(2) with PAA sterilant, more than a 6 log(10) reduction was observed for EMC viruses. Biochemical and biomechanical evaluations showed that the SC-CO(2) treatment with PAA sterilant did not cause significant changes in porcine acellular matrix's susceptibility to collagenase digestion, tensile or tear strength, indicating limited alteration of the tissue structure following SC-CO(2) sterilization.

Effects of e-beam radiation, storage, and hydration on osteoinductivity of DBM/AM composite.

E-beam irradiation is often used to sterilize medical devices including demineralized bone matrix (DBM) products. In this study, the effect of e-beam on osteoinductivity of a DBM product in hydrous and anhydrous configurations has been evaluated at 0-, 6- and 12-month ambient storage using a nude rat muscle pouch model. The thermal and structural stabilities of DBM and acellular dermal matrix (AM) composites were analyzed using differential scanning calorimetry (DSC) and trypsin digestion assay. Both hydrous and anhydrous DBM/AM composites exhibited osteoinductivity after e-beam irradiation of 15 kGy. After 12-month ambient storage, the osteoinductivity of hydrous DBM/AM was diminished, whereas the anhydrous DBM/AM retained its osteoinductive potential. However, the DSC and trypsin analysis revealed that the DBM in anhydrous DBM/AM was more vulnerable to damage from e-beam irradiation than its hydrous counterpart. This study has found that although the anhydrous DBM has more structural damage than hydrous DBM from e-beam irradiation, it has retained its osteoinductivity better after 1-year ambient storage.

Retention of structural and biochemical integrity in a biological mesh supports tissue remodeling in a primate abdominal wall model.

AIM: Suboptimal clinical outcome following the implantation of porcine-derived tissue matrices may be due to the method of processing the material to achieve an acellular graft and to reduce the immune response to xenogeneic epitopes. The ability to produce a porcine-based graft material that retains the structural integrity of the extracellular matrix and minimizes the potential antigenic response to the galactose-alpha(1,3)-galactose terminal disaccharide (alpha-Gal) may allow the scaffold to support regeneration of native tissue. MATERIALS & METHODS: Porcine dermal tissue was processed to remove cells and DNA, and minimize the presence of alpha-Gal via specific enzymatic cleavage. In vivo performance was evaluated by implantation into the abdominal wall of an Old World primate exisional repair model. Grafts were explanted at 0.5, 1, 3 and 6 months and assessed for cellular repopulation and vascularization, for localized immune response by presence of T cells, B cells and macrophages, and systemic immune response by anti-alpha-Gal IgG by ELISA. RESULTS: Animals tolerated implants well and exhibited no clinical signs of inflammation, laxity, hernia or visceral tissue attachment. Histological evaluation revealed marked host fibroblast repopulation and neoangiogenesis as early as 2 weeks postimplant. Cellular repopulation and maturation of vascular structures reached a plateau at 3 months. Immunological evaluation of immune cell infiltration demonstrated an early, mixed cellular inflammatory response at 2 weeks. This cellular immune response was transient and diminished to baseline levels by 3 months postimplant. CONCLUSION: The combination of a nondamaging process, successful removal of cells, and reduction of the xenogeneic alpha-Gal antigens from the porcine dermal matrix, while maintaining an intact extracellular matrix, is critical to its ability to remodel and integrate into host tissue, leading to its overall acceptance.

Host response to implanted porcine-derived biologic materials in a primate model of abdominal wall repair.

Three commercially available porcine-derived biologic meshes were implanted in an Old World primate abdominal wall resection repair model to compare biological outcome as a predictor of clinical efficacy. Tissues were explanted over a 6-month period and evaluated for gross pathology, wound healing strength, mesenchymal cellular repopulation, vascularity, and immune response. In vivo functional outcomes were correlated with in vitro profile for each material. Small intestinal submucosa-based implants demonstrated scar tissue formation and contraction, coincident with mesh pleating, and were characterized by immediate and significant cellular and humoral inflammatory responses. Porcine dermal-based grafts demonstrated significant graft pleating, minimal integration, and an absence of cellular repopulation and vascularization. However, a significant cellular immune response surrounded the grafts, coincident with poor initial wound healing strengths. In vivo observations for the three porcine-derived mesh products correlated with individual in vitro profiles, indicating an absence of characteristic biochemical markers and structural integrity. This correlation suggests that in vivo results observed for these mesh products are a direct consequence of specific manufacturing processes that yield modified collagen matrices. The resulting loss of biological and structural integrity elicits a foreign body response while hindering normal healing and tissue integration.

Effects of gamma-irradiation, storage and hydration on osteoinductivity of DBM and DBM/AM composite.

This study investigated the effect of gamma-irradiation dose, irradiation temperature, hydration and storage condition on osteoinductivity of demineralized bone matrix (DBM) and demineralized bone matrix/acellular dermal matrix (DBM/AM) composite. DBM and DBM/AM in dry and hydrated form were treated with gamma-irradiation of 15-40 kGy at ambient or low temperature (-40 degrees C approximately -70 degrees C) and then stored at ambient condition for 6 months. The athymic rat muscle implant model was used to evaluate the osteoinductive potential of the DBM and DBM/AM composites. Histological and alkaline phosphatase (ALPase) activity assessments were carried out at 28 days after implantation to determine the new bone formation and ALPase activity. Both histological and ALPase activity analysis showed that the osteoinductivity of DBM decreased with the increase of gamma-irradiation dose at ambient temperature, whereas no decrease occurred when treated with gamma-irradiation at low temperature. However, the hydrated DBM showed diminishing osteoinductivity after 6-month storage at ambient condition, whereas the DBM in dry form retained their osteoinductivity after the 6-months storage. The findings in this study indicate that DBM and DBM/AM composites could retain their osteoinductivity when they are in dry configuration and are irradiated at low temperature (-40 degrees C approximately -70 degrees C) using the custom-made cold gamma-irradiation system.

Evaluation of bone regeneration at critical-sized calvarial defect by DBM/AM composite.

This study investigated the bone-regenerative potential of a demineralized bone and acellular matrix (DBM/AM) composite (AlloCraft DBM) in comparison with autologous bone using an in vivo model. Critical-sized calvarial defects (5 mm) were created in athymic rats. The defects were grafted with either the DBM/AM composite or the acellular human dermal matrix (AM), and compared with the defects filled with autologous bone (positive control) and the empty defect (negative control). Histological and radiographic assessments were carried out at 4 and 8 weeks after surgery to determine the biological healing, the amount and type of new bone formation and the percentage of new bone filled in the critical defects. At 4 weeks, DBM/AM composite group had the highest percentage of the defect filled with new bone (84%), which was significantly greater than autologous bone (62%), AM (41%), and untreated control (32%) groups. At 8 weeks, the DBM/AM continued to have the highest percentage of the defect filled with new bone (91%). The autologous bone group increased the percentage of bone fill to 83%. The defects either filled with AM or left untreated still had less of the defect filled with new bone, 57% and 33%, respectively. The total healing of defects grafted with DBM/AM was comparable with autologous bone group at 8 weeks. The results demonstrated that the DBM/AM composite promoted new bone formation more rapidly than autologous bone at calvarial defect in athymic rats. The study supports that DBM/AM is a potential substitute of autologous bone for bone repair.

Evaluation of DBM/AM composite as a graft substitute for posterolateral lumbar fusion.

Demineralized bone matrix (DBM) has been investigated as a bone graft substitute for spinal fusion with less morbidity. Various carriers have been added to DBM to enhance its handling characteristics. This study investigates the spinal fusion induced by a composite of DBM and acellular dermal matrix (AM) in comparison with autologous bone in an athymic rat spinal fusion model. Single-level intertransverse process fusions were performed in 60 athymic nude rats grafted with 2 mL/kg of DBM/AM composite, AM alone, or autologous bone. Fusion was assessed at 6 weeks by radiography, manual palpation, and histology. At 6 weeks, 70% of the animals from the DBM/AM composite group exhibited complete spine fusion, whereas 35% from the autologous bone group and 20% from AM group showed bridging with some gaps. The DBM/AM composite induced a significantly higher fusion rate than both the autologous bone and AM groups (p < 0.001) in all measured parameters. The current study demonstrated that using DBM/AM composite can have more robust fusion than autologous bone at 6 weeks in an athymic rat spinal fusion model.

Flow-cytometric screening of platelet antibodies with previously frozen cells.

UNLABELLED: In this study the flow-cytometric crossmatch results were compared between fresh cells and cells processed by various cryopreservation and storage methods. Platelets from healthy donors were incubated with 12 sera containing platelet reactive antibodies as well as with 62 control sera from blood donors. Direct comparisons were made between fresh platelets and platelets after freezing at -28 degrees C, -40 degrees C and -80 degrees C and in liquid nitrogen, using 6% DMSO as cryoprotectant. In addition, the effects of using controlled-rate freezing were evaluated. Finally we evaluated the application of the cryoprotectant ThromboSol. The best results were obtained after cryopreservation of the platelets with ThromboSol at -80 degrees C with controlled cooling rates. Using ThromboSol cryopreserved platelets, the sensitivity for the detection of incompatible platelets was 100% and the specificity was 97.1%, using the previous results obtained with flow-cytometry, MAIPA and LCT as a reference. CONCLUSION: Platelets can be frozen using ThromboSol as the cryoprotectant, with controlled rate freezing and storage at -80 degrees C for the screening of platelet antibodies and for flow-cytometric crossmatch procedures. This system yields a reproducible and logistically simple method for platelet crossmatching that yields results superior to fresh cells and can be easily incorporated into standard clinical laboratory practices.

T-complex polypeptide-1 interacts with the erythrocyte cytoskeleton in response to elevated temperatures.

Chaperonins are double ring complexes composed of highly conserved 60-kDa protein subunits that are divided into two subgroups. Group II chaperonins are found in archaea and the cytoplasm of eukarya and are believed to function like other chaperonins as part of a protein folding system. We report here that human erythrocytes contain the group II chaperonin T-complex polypeptide 1 (TCP-1) and that this complex translocates from the cytoplasm to the cytoskeleton in response to heat treatment in the absence of overt cell damage. Identification as TCP-1 was determined by immunodetection for TCP-1alpha and corroborated by mass spectroscopy peptide sequencing. Direct visualization by immunofluorescence confirmed peripherally localized TCP-1 in response to heat treatment. Temperatures ranging from 37-50 degrees C were demonstrated to have distinct kinetic profiles of induced translocation. Heat-induced binding was shown by Triton shell analysis to be specifically associated with the cytoskeletal proteins. Furthermore, the binding was reversible following removal of the stimulatory condition. A stabilizing process is hypothesized based on the known interactions of chaperonins.

Biochemical stabilization enhances red blood cell recovery and stability following cryopreservation.

Glycerolized red blood cells (RBC) are approved for long-term cryopreservation. However, the need to remove the glycerol cryoprotectant prior to transfusion has limited the usefulness of this cryopreservation method. This report describes using non-cryoprotectant biochemical stabilization techniques to substitute for the standard glycerol cryoprotectant. The glycerolized RBC method was compared to a newly developed LC-V method that combines transfusable cryoprotectants (hydroxyethyl starch and dextran) and specific non-cryoprotectant biochemical stabilizers (nicotinamide, nifedipine, and flurbiprofen). Results demonstrate that the biochemical stabilizers significantly reduce cryopreservation-induced hemolysis compared to cryopreservation in their absence and that thaw hemolysis levels approach those of standard 40% (w/v) glycerolized RBC (3.1+/-0.2% for 40% glycerol compared to 8.7+/-0.9% for the LC-V protocol). Furthermore, LC-V cryopreserved RBC exhibit a significantly enhanced post-thaw stability compared to glycerolized RBC as determined by osmotic fragility index (0.557+/-0.034 for 40% glycerol compared to 0.478+/-0.016 for the LC-V protocol). Analysis of biochemically stabilized RBC proteins revealed a transient translocation of carbonic anhydrase to the membrane fraction. However, the enhanced RBC recovery and stability could not be attributed to this event. Finally, DSC analysis demonstrated that the biochemical stabilizers of the LC-V process were not functioning as surrogate cryoprotectants in that they did not affect the quantity or quality of ice formed. Overall, this work demonstrates that cryopreservation-induced RBC damage may be corrected or prevented through specific biochemical stabilization and represents a significant step toward a directly transfusable cryopreserved RBC product.

Safety and efficacy of transfusions of autologous cryopreserved platelets derived from recombinant human thrombopoietin to support chemotherapy-associated severe thrombocytopenia: a randomised cross-over study.

BACKGROUND: The increasing demand for platelet products, and concern over the transfusion-associated risks of alloimmunisation and infections, have motivated a search for improved methods aimed at keeping exposure to donor antigens to a minimum. Transfusion of thrombopoietin-derived autologous platelets might provide an alternative strategy. We aimed to compare the safety and efficacy of this strategy with that of transfusion with fresh allogeneic platelets in patients with severe chemotherapy-induced thrombocytopenia. METHODS: 20 patients with gynaecological malignancies were treated with two doses of 1.2 microg/kg recombinant human thrombopoietin. From day 12, we aimed to collect 50 units of platelets from these patients by plateletpheresis. Harvested platelets were cryopreserved in ThromboSol and 2% dimethyl sulfoxide (DMSO) for use in subsequent autologous transfusions. Patients then received carboplatin for up to six cycles. Patients were randomly assigned to group A (n=10), which received allogeneic fresh platelets at the first instance of severe thrombocytopenia (platelet count <15,000/microL) and then autologous cryopreserved platelets at the next, or to group B (n=10), which received first autologous and then allogeneic platelets. In subsequent cycles, all patients received autologous platelets while available. The primary endpoint was platelet count increment corrected for the number of platelets transfused and the patients' body-surface area. Analysis was by intention to treat. FINDINGS: Treatment with recombinant human thrombopoietin significantly increased platelet count (median 2.3-fold [range 1.5-3.3], p<0.0001) in all but one patient in group A. The median number of platelets collected per patient was 53 units (14-66) in two collections (one to three). There was no significant difference in the corrected platelet count increments (CCIs) between the 19 paired transfusions of cryopreserved autologous platelets and fresh allogeneic platelets (median 1-h CCI 15.7 vs 19.8, p=0.398; median 24-h CCI 13.0 vs 18.1, p=0.398). 14 of the 19 patients had a good response (1-h CCI >7.5) to their first transfusion of allogeneic platelets. By contrast, all patients had a good response to their first transfusion of autologous platelets (p=0.063). Moreover, no significant decrease in the CCIs (p=0.405) was seen over six cycles after autologous platelet transfusions (n=63). No transfusion reactions or any serious adverse event was recorded during autologous platelet transfusions. INTERPRETATION: Recombinant human thrombopoietin facilitated collection of multiple units of platelets, which could be cryopreserved and reinfused to counteract severe thrombocytopenia during multicycle chemotherapy. Transfusion of autologous cryopreserved platelets derived from recombinant human thrombopoietin can provide a viable strategy to minimise the risks of allogeneic platelet transfusions and provide a long-lasting supply of platelet support.